JMI CONTRACTORS LLC v. MEDELLIN (2026)

While working on a construction site as an independent contractor, Jose Medellin fell off a roof and sustained serious injuries. He sued the general contractor, JMI Contractors, LLC, alleging negligent activity and premises liability. A jury sided with Medellin on both theories. It awarded him substantial damages, the trial court entered judgment on the verdict, and the court of appeals affirmed.

This case implicates two questions that frequently recur in our jurisprudence. The first involves whether a workplace injury gives rise to a cause of action for negligent activity or premises liability—an issue that often proves difficult to assess. The second asks, in the context of premises-liability claims, whether independent contractors like Medellin can recover for injuries caused by open and obvious dangers.

Today, we examine our precedents carefully and follow where they lead. We first conclude that Medellin's claim sounds in premises liability, not negligent activity, because his injury stems from nonfeasance, not malfeasance. The record here does not support an independent claim for negligent activity against JMI.

We further conclude that Medellin's premises-liability claim fails as a matter of law. In two previous decisions, we have suggested, but never squarely held, that independent contractors cannot recover in premises liability for an injury resulting from an open and obvious danger. Today, we elevate those suggestions to a holding. As an independent contractor, Medellin agreed to work on a roofing project, which necessarily presented the open and obvious risk of a fall. Medellin cannot maintain a claim for premises liability when that open and obvious danger caused his injury.

Accordingly, we reverse the judgment below and render a take-nothing judgment in favor of JMI.

I

A

In March 2018, JMI, a general contractor, hired independent contractor Metal Roof & TPO Specialist, LLC, for a roofing project at the Oaks on Bandera Apartments in San Antonio. Metal Roof in turn retained Medellin to assist. Medellin works in construction as an independent contractor. There is no allegation that he was at any time an employee of either JMI or Metal Roof. The record indicates that Medellin had substantial experience with roofing projects, and he had worked with Metal Roof in the past.

When Medellin arrived at the job site, he learned he would be working with Raul Rodriguez and Reybel Rodriguez. Raul was an employee of JMI, and he served as supervisor of the job site. Reybel, Raul's nephew, was an independent contractor.

The project involved stretching a patch of a rubberized roofing membrane over part of the roof. This task required multiple workers, moving together and pulling the membrane in a coordinated fashion. Unfortunately, while pulling the roll, Medellin lost track of the roof's edge. He stepped off, fell approximately 30 feet to the ground below, and sustained serious injuries.

B

This lawsuit followed. Medellin asserted against JMI various theories of liability stemming from the fall. The matter proceeded to a jury trial. Relevant here, the jury considered two claims: negligent activity and premises liability. The jury found JMI liable for both. It awarded Medellin over $3.3 million in compensatory damages and an additional $1 million in exemplary damages.

JMI appealed, raising the two issues that form the basis of our decision today. First, JMI argued that Medellin's negligent-activity claim fails because Medellin's injury arises from nonfeasance—an unguarded roof edge. Second, JMI argued that the necessary-use exception should not apply to Medellin because he is an independent contractor.¹

The court of appeals initially sided with JMI and ordered a new trial based on an issue not relevant to our decision today. JMI Contractors, LLC v. Medellin, 2023 WL 4217036, at *7-9 (Tex. App.—San Antonio June 28, 2023, no pet.), withdrawn & superseded on reh'g, 716 S.W.3d 783, 794 (Tex. App.—San Antonio 2024). But on Medellin's motion for rehearing, the court of appeals withdrew its prior opinion and substituted a new opinion affirming the trial court's judgment. 716 S.W.3d at 794.

As to premises liability, the court of appeals noted that our Court has previously “expressed doubt” that independent contractors can avail themselves of the necessary-use exception because we generally expect independent contractors to “take into account any open and obvious premises defects in deciding how the work should be done.” Id. at 799 (quoting SandRidge Energy, Inc. v. Barfield, 642 S.W.3d 560, 568 (Tex. 2022)). Nevertheless, the court of appeals held that the exception applied “because the evidence shows JMI controlled the safety requirements of the jobsite” and that “JMI should have anticipated Medellin ․ was unable to take the safety measures required to avoid the risk of falling off a flat roof.” Id. at 799-800.

The court of appeals likewise held that a negligent-activity theory supported the trial court's judgment because “JMI actually controlled the safety requirements and manner of Medellin's work on the jobsite and therefore owed Medellin a duty of care.” Id. at 797. The court offered little analysis on whether any negligent activity was properly attributed to JMI. See id. at 797-800.

We granted JMI's petition for review.

II

We begin with a forensic review of our precedents on negligent activity and premises liability, two “separate and distinct theories of recovery, requiring plaintiffs to prove different, albeit similar, elements to secure judgment in their favor.” United Scaffolding, Inc. v. Levine, 537 S.W.3d 463, 471 (Tex. 2017). The development of our doctrine sheds light on the state of the law that produces our decision today.

A

The relevant caselaw begins in 1948, with Houston National Bank v. Adair. 146 Tex. 387, 207 S.W.2d 374 (1948). A bank customer slipped and fell on a smooth marble stairway. Id. at 374. We rejected her liability theory because she could see that the stairs were slippery, yet traversed them anyway. Id. at 375. We concluded that a premises owner has no duty to protect invitees from dangerous conditions that are “open, visible and obvious to any reasonably careful person.” Id. at 376. This established the foundational no-duty rule for open and obvious conditions: if the invitee can perceive the danger as easily as the owner, there is no duty to warn or ameliorate. Id.

The difficulty, however, is that many real-world injuries are caused by a confluence of factors—some involving a dangerous condition in the premises themselves, and some involving dangerous activities on otherwise safe premises. See, e.g., E. Tex. Theatres, Inc. v. Rutledge, 453 S.W.2d 466 (Tex. 1970) (a moviegoer injured by a bottle thrown from the balcony implicated negligent activity, not premises liability, even though the elevated balcony contributed to the injury); Abalos v. Oil Dev. Co. of Tex., 544 S.W.2d 627 (Tex. 1976) (injury at an oilfield operating pump implicated negligent activity, not premises liability, despite oilfield dangers). As we have observed, “[a]t some point, almost every artificial condition can be said to have been created by an activity.” Keetch v. Kroger Co., 845 S.W.2d 262, 264 (Tex. 1992). In these situations, can a plaintiff proceed under both theories, or just one—and if just one, which one?

We explored that question in our 1992 decision in Keetch, which presented a grocery store slip-and-fall. Id. at 263. The plaintiff argued that the slick patch resulted from the overuse of a leaf-shining product that had been sprayed on flowers but dripped onto the floor, forming “a waxy type substance.” Id. at 264. We considered whether the injury sounded in the negligent activity of using too much shining spray, or the premises defect of a slippery floor. Id. We chose the latter and explained that “[r]ecovery on a negligent activity theory requires that the person have been injured by or as a contemporaneous result of the activity itself rather than by a condition created by the activity.” Id. The plaintiff “may have been injured by a condition created by the spraying but she was not injured by the activity of spraying.” Id.

We reemphasized the contemporaneous-activity rule a few years later in Clayton W. Williams, Jr., Inc. v. Olivo, 952 S.W.2d 523, 527 (Tex. 1997). The plaintiff there worked as a floor hand for an independent contractor hired to drill a well. Id. at 526. He slipped off a pipe rack and suffered an injury when he landed on thread protectors left on the ground by the previous shift. Id. at 526-27. Was leaving the thread protectors lying around an ongoing negligent act, or were the mislaid thread protectors a dangerous condition on the property? Again, we chose the latter. As we summed up, “[t]his is not a negligent-activity case because [the plaintiff] alleges that he was injured by thread protectors previously left on the ground, not as a contemporaneous result of someone's negligence.” Id. at 527 (citing Keetch, 845 S.W.2d at 264).

The next year brought us Timberwalk Apartments, Partners, Inc. v. Cain, 972 S.W.2d 749 (Tex. 1998). The plaintiff was assaulted by an intruder who entered her apartment through a sliding glass door. Id. at 751. The apartment building provided little security; there were no working access gates, no alarm systems, no guards, and no pin locks for the sliding doors. Id. The plaintiff sued the landlord over its security measures. Id. Once again, we considered “whether [the] plaintiff alleges a negligent activity or a premises defect.” Id. And once again, we chose the latter. Even though the plaintiff's injury resulted from a contemporaneous activity—the intruder's assault—we concluded that the landlord's liability sounded in a premises defect. Id. at 753. It was the inherently unsafe and persistent condition of the premises, in other words, that controlled the analysis with respect to the landowner.

In 2010, Del Lago Partners, Inc. v. Smith again brought us the confluence of violent criminal acts and premises liability. 307 S.W.3d 762, 775-76 (Tex. 2010). A vicious barroom brawl broke out at a resort. Id. at 766. The staff kept serving drinks even as the tensions escalated, and no one called security until after the melee had injured the plaintiff. Id. at 771. Leaning on Timberwalk, we analyzed the plaintiff's claim under a premises-liability theory. Id. at 775-77. The dangerous condition, we reasoned, was the extended period of escalating hostility among intoxicated patrons—an unreasonable risk of harm on the premises that the owner observed but did nothing to remedy. Id. at 770, 776-77.

Justice Wainwright dissented. He argued that nothing about the bar itself posed a danger. Rather, the danger arose entirely from the contemporaneous conduct of patrons and employees. Id. at 787 (Wainwright, J., dissenting). Partygoers drank, belligerents fought, the staff kept serving, and no one intervened until it was too late. Id. But the Court “disagree[d]” and held “the case was properly submitted on a premises-liability theory,” largely because “[w]e have repeatedly treated cases involving claims of inadequate security as premises-liability cases.” Id. at 775, 776 (citing Timberwalk, 972 S.W.2d at 753). To the Court, the “failure to remedy an unreasonably dangerous condition for ninety minutes and failure to react promptly once the fight started” was a problem of nonfeasance, not malfeasance. Id. And nonfeasance gives rise to premises liability even when contemporaneous activity plays a key role in producing the injury. Id.

That brings us to the 2015–17 period, when we issued four decisions critical to shaping the doctrine as it stands today: Austin v. Kroger Texas, L.P., 465 S.W.3d 193 (Tex. 2015); Occidental Chemical Corp. v. Jenkins, 478 S.W.3d 640 (Tex. 2016); Sampson v. University of Texas at Austin, 500 S.W.3d 380 (Tex. 2016); and United Scaffolding, 537 S.W.3d 463. All four considered whether a particular injury sounded in negligent activity or premises liability, and in all four, we held that premises liability applied.

Austin involved an employee–plaintiff who slipped and fell while mopping the restroom floor of a grocery store. 465 S.W.3d at 198. The employee argued in part that he was engaged in a negligent activity because he should have been using a special absorbent powder to clean up the spill—not a mop—under the employer's safety handbook. Id. We rejected that argument. Id. at 216. Even though the alleged negligent activity was undoubtedly contemporaneous, we emphasized that the real problem was a condition of the premises, which implicates “nonfeasance, and not contemporaneous negligent activity.” Id. Accordingly, “it could not give rise to a negligent-activity claim.” Id.

The plaintiff in Occidental Chemical was injured at a chemical plant when a chemical system malfunctioned and sprayed acid in his face. 478 S.W.3d at 642. We reasoned that the defendant had designed and installed the faulty chemical system several years earlier, and at the time of the injury, the system represented “a dangerous condition on its property,” and not “a contemporaneous, negligent activity.” Id. at 644-45.

The plaintiff in Sampson was injured when he tripped over an unsecured extension cord strung across a walkway at mid-shin height. 500 S.W.3d at 383. The plaintiff claimed that the failure to secure the cord or tape it to the ground was a negligent act. Id. We disagreed: the cord's static placement was a premises defect. Id. at 390-91.

In United Scaffolding, the plaintiff was injured when he slipped on unsecured plywood covering a hole in a scaffold built by the defendant the previous week. 537 S.W.3d at 467. The defendant was not present on site at the time of the plaintiff's fall. Id. The plaintiff alleged that the unsafe placement of the plywood was a negligent act. Id. at 468. We squarely rejected a negligent-activity theory and held that the plaintiff's claim sounded in premises liability. Id. at 479. We noted each of the cases discussed above, and we identified their core principles. First, we recognized that negligent activity and premises liability “are separate and distinct theories of recovery, requiring plaintiffs to prove different, albeit similar, elements to secure judgment in their favor.” Id. at 471. They “are based on independent theories of recovery,” and as a result, “they are not interchangeable.” Id. That is, a plaintiff's single undivided injury may sound in premises liability or negligent activity—but not both. See id.; see also Occidental Chem., 478 S.W.3d at 644 (“[A] person injured on another's property may have either a negligence claim or a premises-liability claim against the property owner.” (emphases added)). We then reiterated the nonfeasance-versus-malfeasance distinction from Del Lago. United Scaffolding, 537 S.W.3d at 471 (citing Del Lago, 307 S.W.3d at 776). And we “recognized that ‘slip/trip-and-fall cases have consistently been treated as premises defect causes of action.’ ” Id. (citing Sampson, 500 S.W.3d at 391).

As the above discussion demonstrates, we have time and again addressed the distinction between premises liability and negligent activity. We distill below, in Part II.C, the core principles these cases embody. But first, we must detour to consider an important doctrinal wrinkle—one the parties here brief extensively.

B

That wrinkle is Lee Lewis Construction, Inc. v. Harrison, 70 S.W.3d 778 (Tex. 2001), which we decided in 2001, many years before our more recent pronouncements in Del Lago, Austin, Occidental Chemical, Sampson, and United Scaffolding. The decedent was an employee of a subcontractor working at substantial elevation on the upper floors of a construction project. Lee Lewis, 70 S.W.3d at 782. The general contractor had taken full control over the means of fall protection and had dictated the use of bosun's chairs—a wooden board suspended from the roof by a single rope—without independent lifelines. Id. at 784. In so doing, the general contractor usurped the independent contractor's traditional authority to provide for his own safety measures and created a danger. Id. While using a bosun's chair to perform exterior window work on the tenth floor, the decedent fell to his death. Id. at 781-82.

As it came to us, Lee Lewis was framed very differently than the cases discussed above, all of which explored the distinctions between negligent activity and premises liability. In Lee Lewis, it appears that no party argued that the plaintiff's claim sounded in premises liability. As such, Lee Lewis did not address any argument about whether a particular type of injury sounds in premises liability or negligent activity. See Cooper Indus., Inc. v. Aviall Servs., Inc., 543 U.S. 157, 170, 125 S.Ct. 577, 160 L.Ed.2d 548 (2004) (“Questions which merely lurk in the record, neither brought to the attention of the court nor ruled upon, are not to be considered as having been so decided as to constitute precedents.” (quoting Webster v. Fall, 266 U.S. 507, 511, 45 S.Ct. 148, 69 L.Ed. 411 (1925))).

Instead, Lee Lewis considered the circumstances in which those who retain control over an independent contractor's work may be liable for injuries suffered in the course of that work. 70 S.W.3d at 783-84. We observed that the general contractor had not only retained full control over the fall-prevention measures of the subcontractors' employees but had in fact created a danger by insisting on the use of bosun's chairs without lifelines. See id. at 784; see also Dow Chem. Co. v. Bright, 89 S.W.3d 602, 609 (Tex. 2002). Under these circumstances, we concluded that the defendant undertook and breached a duty of care. Lee Lewis, 70 S.W.3d at 782-84.

In the years following Lee Lewis, we issued two important decisions refining its application. The first was Dow Chemical, 89 S.W.3d 602. There, an employee of an independent contractor was injured by an unsecured pipe during construction work. Id. at 605. He sued the landowner, invoking Lee Lewis to argue that the landowner had taken on a duty of care through its control over the premises. Id. at 609. We disagreed and rendered judgment for the landowner because none of the plaintiff's indicia of control—safe work permit systems, safety manuals, safety representatives, and so on—sufficiently established control. Id. at 607-11. We held that merely promulgating safety policies, having safety personnel on site, or retaining a general right to stop unsafe work does not amount to the kind of control over the means, methods, and details of work needed to impose a duty. Id. And (as will become relevant here) we added that the “failure to implement” a plaintiff's proposed “additional safety” systems “is not actual control.” Id. at 609.

Then, in General Electric Co. v. Moritz, we encountered a plaintiff who, while working as an independent contractor, routinely used a loading-dock ramp at a warehouse to load and unload a truck. 257 S.W.3d 211, 213 (Tex. 2008). The ramp had no guardrails. Id. While securing a load, the plaintiff fell from the ramp and suffered an injury. Id. at 213-14. He sued the landowner, asserting negligent activity and premises defect. Id. at 214. As to premises liability, the plaintiff argued the defendants should have installed guardrails on the ramp. Id. at 215. We dismissed such a theory, holding that the defendants owed no premises-liability duty to the plaintiff regarding a hazard that was not concealed. Id.²

As to negligent activity, the plaintiff invoked Lee Lewis to argue that the defendant generally controlled the means of loading the truck—the activity that caused the injury. We squarely rejected that theory and rendered judgment for the defendant. Id. at 214-15 (distinguishing between control over where the truck was loaded as opposed to how that load was secured). Importantly, we discussed Lee Lewis as a negligent-activity case, and a negligent-activity case only. Id. In that discussion, we held that “it is not enough to show that the defendant controlled one aspect of [the plaintiff's] activities if his injury arose from another.” Id. at 214 (emphases added). A generalized control over working conditions, in other words, is insufficient. See id.

Any reliance on Lee Lewis must necessarily be informed by the analysis in Dow Chemical and Moritz. In light of those subsequent decisions, we believe Lee Lewis is best understood to apply in the narrow set of circumstances present in that case. Those circumstances include the general contractor's decision to fully control the particular plaintiff's activity in a way that creates a heightened danger that in turn causes that plaintiff's injury. As we recount above, the Lee Lewis defendant required the use of a bosun's chair without an independent lifeline—a dangerous device that led to the decedent's death. That is the scenario that independently gives rise to liability notwithstanding a dangerous condition of the premises.

Although Moritz treated Lee Lewis as a negligent-activity case, we have never squarely explained how the Lee Lewis retained-control framework fits with our later cases indicating that a single injury generally sounds in premises liability, or negligent activity, but not both. Lee Lewis itself never contemplated the antecedent question of how to properly characterize that claim in the first place—whether it is one for premises liability or negligent activity. It was not until 16 years later, when we decided United Scaffolding, that we insisted lower courts channel the claim into its appropriate lane depending on whether it sounds in nonfeasance or malfeasance. 537 S.W.3d at 471.

We adhere to what we said in United Scaffolding: whether a single injury sounds in negligent activity or premises liability depends on whether it implicates malfeasance or nonfeasance, two distinct concepts. Id. We have never categorically rejected the possibility that unusual circumstances might involve both malfeasance and nonfeasance and thus implicate both negligent activity and premises liability. See Austin, 465 S.W.3d at 216 n.22 (declining to resolve “whether a single injury could give rise to both a premises-liability claim and a negligent-activity claim”). We do not delve into that abstract question today, and we leave open the possibility that even if a claim is properly understood to sound in premises liability, there may be rare instances where a plaintiff may assert a negligent-activity claim as well—whether under the Lee Lewis retained-control framework or otherwise.³

C

Summing up, our precedents establish a handful of core propositions that guide our decision today.

First, whether an injury implicates premises liability or negligent activity presents a question of law that should be resolved by the court. United Scaffolding, 537 S.W.3d at 469 (“Whether the condition that allegedly caused the plaintiff's injury is a premises defect is a legal question, which we review de novo.”). A plaintiff's artful pleading does not turn a negligent-activity claim into a premises-liability claim, nor vice versa. Id. at 480 (“[I]f a claim is properly determined to be one for premises defect, a plaintiff cannot circumvent the true nature of the claim by pleading it as general negligence.” (quoting Sampson, 500 S.W.3d at 389)); see also Occidental Chem., 478 S.W.3d at 648 (clarifying that pleadings cannot overcome premises-liability character of claim). Indeed, several of our cases were pleaded as negligent-activity cases, yet we did not hesitate to set aside the plaintiff's preferred framing and decide the case under the premises-liability framework. See H.E. Butt Grocery Co. v. Warner, 845 S.W.2d 258, 259 (Tex. 1992) (“Although Warner maintains that she has been denied her alternative, negligent failure to pre-bag chicken theory of liability, it is undisputed that she was injured by a condition of the premises.”); Keetch, 845 S.W.2d at 264 (rejecting alleged error in “not submitting [the plaintiff's] negligent activity theory” to the jury because there “was no ongoing activity when [the plaintiff] was injured”); United Scaffolding, 537 S.W.3d at 483 (holding that it was improper to submit the plaintiff's claim under a general negligence theory because, “properly characterized,” his claim was one for premises liability).

Second, the core distinction between the two theories is one of malfeasance versus nonfeasance. Negligent-activity claims are grounded in malfeasance—“affirmative, contemporaneous conduct by the owner that caused the injury.” Del Lago, 307 S.W.3d at 776 (citing Timberwalk, 972 S.W.2d at 753). Premises-liability claims are grounded in nonfeasance—“failure to take measures to make the property safe.” Id. In other words, negligent-activity liability requires that the plaintiff be injured by or as a contemporaneous result of the negligent activity itself rather than a condition. Premises-defect liability, on the other hand, “is based on the property itself being unsafe.” State v. Shumake, 199 S.W.3d 279, 284 (Tex. 2006).

Third, a negligent-activity claim requires contemporaneous action. Even a 30-minute gap between the chemical spray and the slip, as in Keetch, is not enough to sustain a negligent-activity theory. 845 S.W.2d at 264. The same goes for dangerous equipment left lying around—as in Olivo—and for negligently built structures—as in United Scaffolding. See also Sampson, 500 S.W.3d at 391 (plaintiff who tripped over improperly secured extension cord had only premises-liability claim); Occidental Chem., 478 S.W.3d at 648 (rejecting idea that property owner can be subjected to premises liability and ordinary negligence, even when the owner creates the negligent condition).

Fourth, when an injury results from a confluence of factors involving human activity and the condition of the premises, we decide which controls by analyzing the essence or gravamen of that particular claim against that particular defendant. United Scaffolding, 537 S.W.3d at 480; see also Warner, 845 S.W.2d at 259 (“[W]e hold that [the plaintiff's] First Amended Original Petition, fairly read, alleged only a claim based upon premises liability.”); Del Lago, 307 S.W.3d at 776 (claim “properly tried and submitted as a premises-liability case” because the plaintiff “primarily complained” of nonfeasance). As United Scaffolding put it, a plaintiff “cannot transform the true nature of a claim” with artful pleading or by submitting a different theory to the jury. 537 S.W.3d at 480. Put another way, we ask whether the negligent activity can be defined without reference to the condition of the premises. Telling someone to carry a book is not dangerous, but telling someone to carry a book through a minefield is. When the book carrier is injured by an exploding mine, his claim sounds in premises liability, not negligent activity, because the activity is only dangerous by reason of its relationship to a condition of the premises. On the other hand, directing someone to run with scissors is dangerous everywhere. Cf. Redinger v. Living, Inc., 689 S.W.2d 415, 417 (Tex. 1985). When the plaintiff suffers a stab wound, he may recover for negligent activity.

Finally, to fall within the control framework presented in Lee Lewis, as clarified by Dow Chemical and Moritz, a plaintiff must establish that the defendant retained full control over the plaintiff's activity and increased the risk of injury by creating a danger. See Moritz, 257 S.W.3d at 214-15. In this narrow scenario, liability in negligence may attach.

With those principles in mind, we turn to the circumstances of this dispute.

III

Medellin argues that his injury results from negligent activity, and he asks us to uphold the jury's negligence finding on that basis. The negligent activity, according to Medellin, was the instruction to pull the rubber membrane across the surface of the roof without adequate regard for Medellin's perilous proximity to the roof's edge.

The gravamen of this claim, we hold, sounds in nonfeasance, not malfeasance. At core, Medellin claims that JMI failed to adequately safeguard the roof's edge. Adhering to our directive in United Scaffolding to channel a claim into its appropriate lane without regard to the plaintiff's chosen labels, 537 S.W.3d at 471, we conclude that Medellin's claim against JMI is properly understood to implicate premises liability. We analyze the merits of that claim below, in Part IV.

Before we do so, however, we consider whether Medellin may separately maintain a negligent-activity claim against JMI. As we note above, our precedents leave open the possibility that in unusual circumstances, a plaintiff may assert a distinct claim for negligent activity when the defendant's negligent act independently causes or contributes to an injury. See Austin, 465 S.W.3d at 216 n.22 (“We do not decide here whether a single injury could give rise to both a premises-liability claim and a negligent-activity claim if both the condition of the premises and the contemporary activities of the premises owner proximately cause the injury.”). We need not and do not hold conclusively that a plaintiff could never assert an independent negligent-activity claim in connection with a fall while performing roofing work.

Nevertheless, we conclude that on the record before us, Medellin's negligent-activity claim against JMI fails as a matter of law. No competent evidence ties the alleged independent negligent activity—the instruction to pull the rubber membrane—to JMI. Nor does it establish the “retained control” necessary to establish liability against JMI under the Lee Lewis line of cases. We will explain our reasoning as to both conclusions.

A

The negligent activity Medellin identifies in his briefing is the instruction to pull the rubber membrane near the roof's edge. In concluding that the record does not support liability for this claim, we draw heavily from our decision in Moritz. There, we required the plaintiff to demonstrate the defendant's control over the alleged injury-causing activity—securing a trucking load. Moritz, 257 S.W.3d at 214-15. The plaintiff failed to show that the defendant controlled how he secured the load. Id. To be sure, the defendant may have controlled where the plaintiff loaded the truck, but securing that load was left to the plaintiff. Id. at 215. And since the fall was the result of how (not where) the load was secured, no claim lay against the defendant. Id. On this aspect of that case, our Court was unanimous.

We follow that reasoning here. The record does not tie the alleged injury-causing activity—the instruction to “pull the whole roll”—to JMI. Rather, that instruction came from another independent contractor, Reybel Rodriguez. The record indicates that Reybel was on the other end of the roll and told Medellin to “pull the roll.” So, too, does Medellin's briefing in our Court, which confirms that it was Reybel who gave the instruction at issue. See Resp. Br. at 6.

To be sure, a different individual named Rodriguez—Raul Rodriguez—was also present on site the day of Medellin's fall. Raul Rodriguez was a JMI employee. But Reybel Rodriguez was not; he worked for a different construction company. For Reybel's actions to be attributable to JMI, Medellin would need to demonstrate an employer–employee or other agency relationship between JMI and Reybel. See Exxon Mobil Corp. v. Rincones, 520 S.W.3d 572, 588-90 (Tex. 2017). He has not done so. Although there are jury findings regarding JMI's control over and relationship to Raul, there are no such jury findings regarding Reybel. Thus, as in Moritz, JMI cannot be held liable for the instruction to pull the roll. See 257 S.W.3d at 214 (“[I]t is not enough to show that the defendant controlled one aspect of [the] activities if [the plaintiff's] injury arose from another.”); see also Massage Heights Franchising, LLC v. Hagman, 712 S.W.3d 615, 622 (Tex. 2025) (“[T]he control exercised must relate directly to the injury caused by the negligence.”).

To the extent Medellin argues that the failure to provide adequate safety measures is itself a negligent activity, that argument is foreclosed by our precedents. We squarely held in Del Lago and Moritz that the failure to make a hazardous condition safer sounds in premises liability, not negligent activity. Del Lago, 307 S.W.3d at 775-76; Moritz, 257 S.W.3d at 214-17. We see no basis to revisit that sound principle and overrule our precedents. And we see no additional evidence in this record that would support the imposition of liability for negligent activity.

B

We further hold that Medellin cannot avail himself of the “retained-control” framework set out in Lee Lewis and informed by Dow Chemical and Moritz. To do so, Medellin would have to establish that JMI itself fully controlled Medellin's safety and in so doing created an increased danger. See supra Part II.B. The record here does not bear that out.

As an initial matter, the jury here never found that JMI retained control over Medellin's safety. The trial court asked the jury whether JMI “exercise[d] or retain[ed] some control over the use of fall protection safety measures for the roofing work performed at the Oaks on Bandera on the day of the incident.” The jury answered “Yes.” But that general finding of “some control” as to some employees is insufficient. Our precedents require a particularized finding of control linking the defendant to the plaintiff. E.g., Moritz, 257 S.W.3d at 214-15. That an employer might generally provide some safety measures—e.g., hard hats on a construction site—does not establish retained control.

And to the extent Medellin argues that JMI did not do enough to reduce his risk of injury—by, for example, better guarding the roof edge—such nonfeasance claims sound in premises liability. United Scaffolding, 537 S.W.3d at 471. As we recognized in Dow Chemical just one year after Lee Lewis was decided, the failure to implement a safety procedure “is not actual control.” 89 S.W.3d at 609.⁴

IV

We turn now to premises liability. Generally, premises-liability claims cannot succeed when the danger at issue is open and obvious. See Austin, 465 S.W.3d at 213. Medellin agrees that an unguarded roof edge is an open and obvious danger. His claim thus fails as a matter of law unless he can avail himself of a common-law doctrine we have called the necessary-use exception. This doctrine provides that despite an invitee's awareness of the risk presented by an open and obvious condition, a landowner or occupier may nevertheless be liable if the invitee must necessarily confront the hazardous condition and the owner or occupier “should have anticipated that the invitee is unable to take measures to avoid the risk.” Id. at 208.

On at least two occasions, we have questioned whether the necessary-use exception reaches the circumstances presented here, where an independent contractor seeks to impose liability on a general contractor or landowner in connection with the open and obvious dangers of the jobsite. SandRidge, 642 S.W.3d at 568 (“Our Court has expressed doubt that the necessary-use exception applies to independent contractors.”); Moritz, 257 S.W.3d at 215-16 (“Thus, one who hires an independent contractor generally expects the contractor to take into account any open and obvious premises defects in deciding how the work should be done, what equipment to use in doing it, and whether its workers need any warnings.”). We now take up that question head-on, and we conclude that the necessary-use exception does not reach independent contractors.

A

To properly assess whether the necessary-use exception should apply to independent contractors, we begin by tracing the development of the doctrine.

A plaintiff bringing a premises-liability claim must demonstrate four well-explored elements: (1) the owner's knowledge of the condition, (2) an unreasonable risk of harm, (3) the owner's lack of reasonable care to reduce or eliminate the risk, and (4) proximate causation. Keetch, 845 S.W.2d at 264 (citing Corbin v. Safeway Stores, Inc., 648 S.W.2d 292, 296 (Tex. 1983)). Even if an owner might otherwise be liable under that framework, we have recognized a carveout when the risks posed by the unreasonably dangerous condition were “open or obvious to a person exercising ordinary care.” Hou. Nat'l Bank, 207 S.W.2d at 376. At one point in our history, we referred to this as the “no-duty rule” because it relieved owners of any duty they might otherwise carry to make the condition safe. See generally Halepeska v. Callihan Ints., Inc., 371 S.W.2d 368, 381-84 (Tex. 1963) (discussing the history of the no-duty rule). We now call it the open-and-obvious doctrine.⁵

The original justifications for this doctrine were straightforward. When an invitee is aware of a danger on the premises, he is at least as well positioned as the landowner—perhaps even better positioned—to avoid that danger. So our law incentivizes him to do so, by preventing him from imposing tort liability on the landowner. This general rule supports the commonsense notion that if one knows of a danger, he cannot recover when he fails to avoid it. See id. at 383.

But what if the invitee cannot avoid the danger even though he is aware of it? We confronted that scenario in 1978 in the seminal Parker v. Highland Park, Inc. 565 S.W.2d 512 (Tex. 1978). That case involved a plaintiff who visited her sister's apartment. Id. at 513. The only way to exit the apartment was through a poorly lit and uneven staircase. See id. at 514. The plaintiff knew of this danger and tried to mitigate it with a flashlight. Id. But one night, she tripped, suffered an injury, and sued. Id.

We upheld the judgment in her favor. We recognized that the danger was indeed open and obvious—but unavoidable. The plaintiff had to leave her sister's apartment, and the only way to do so involved confronting a hazard that she could neither make safe nor circumnavigate. Thus was born our “necessary-use exception” to the open-and-obvious doctrine: an invitee may recover in premises liability in connection with an open and obvious defect when (1) the invitee necessarily had to confront the defect, and (2) the landowner should have known the hazard was unavoidable. Austin, 465 S.W.3d at 207.⁶

Several decades later, we confronted a related question: does the necessary-use exception apply to an employee whose very job is to remedy the dangerous condition? That question came to us in Austin v. Kroger Texas, L.P., via certification from the U.S. Court of Appeals for the Fifth Circuit. 465 S.W.3d at 199. On the one hand, an employee tasked with remedying a dangerous condition plainly must encounter that condition—the traditional necessary-use exception. On the other hand, “making landowners liable to employees for such conditions directly disincentivizes employers from hiring employees to remedy such conditions.” Id. at 214.

We sided with the employer. In so doing, we cast Parker as a limited exception—the necessary-use exception—to the general rule that there is no duty to protect an invitee against premises defects that are “open and obvious or otherwise known.” Id. at 207. Effectuating the policy considerations underlying these doctrinal concepts, we concluded that it would make no sense to impose liability on an employer for directing an employee to remedy a known premises defect. Id. at 213. After all, our public policy encourages landowners and employers to “eliminate a dangerous condition,” id. at 214, and holding employers liable for attempting to do so with trained employees would be antithetical to that end. Furthermore, we noted that applying the necessary-use exception to employees directed to encounter the hazard as part of their work would result in the exception swallowing the rule. Id. at 213. The upshot, following Austin, is that an employee cannot avail himself of the necessary-use exception to impose liability when his job requires him to confront the danger at issue.

We have never squarely addressed whether the necessary-use exception reaches independent contractors, but we have twice expressed doubt that it should. First, in Moritz, we held that that the defendant owed no duty to warn of a loading ramp without railings because the hazard was “obviously not [ ] concealed” and because independent contractors are expected “to instruct [their] own employees on the safe means and manner” of performing the job. 257 S.W.3d at 215, 218. We explained that “[p]lacing the duty on an independent contractor to warn its own employees or make safe open and obvious defects ensures that the party with the duty is the one with the ability to carry it out.” Id. at 216. We further emphasized that while the plaintiff had to use the loading dock to load the truck, the defendant did not control how he secured the load. Id. at 217. As a result, “he could not require [the defendant] to keep him safe no matter how he chose to do his own work.” Id.

We reiterated those concerns just four years ago in SandRidge, 642 S.W.3d at 568. Nevertheless, we did not resolve whether the exception applies to independent contractors because the plaintiff had not demonstrated that the landowner would anticipate that the plaintiff's employer was “unable to take measures to avoid the risk.” Id. at 569 (quoting Austin, 465 S.W.3d at 208). There, the plaintiff was a power lineman working for an independent contractor hired to add neutral lines to poles carrying energized lines. Id. at 564. The plaintiff was tasked with removing energized “hot taps” during the work. Id. Because he had a specialized tool for use on energized lines and had been working this job for six months prior to the injury, we held that SandRidge could expect the plaintiff's employer to “take the energized lines into account in instructing and equipping [the plaintiff] to avoid the risk.” Id. at 569. We left open whether independent contractors could ever invoke the necessary-use exception. Id.

B

We now take up JMI's argument that the necessary-use exception does not extend to independent contractors.

1

As an initial matter, Medellin argues that this issue is forfeited because it was not adequately preserved in the trial court. We express no view on the particulars of trial-court preservation here. We instead conclude that because the court of appeals decided the legal question of whether independent contractors may avail themselves of the necessary-use exception, we may review that portion of the judgment below without the need to consider trial-court preservation.⁷

Texas Rule of Appellate Procedure 53.1 authorizes us to “review a court of appeals' final judgment.” See also Tex. Gov't Code § 22.001(a) (“The supreme court has appellate jurisdiction ․ of [a] ․ judgment.”). The final judgment on review here is grounded in the holding that Medellin can avail himself of the necessary-use exception despite his status as an independent contractor. The court of appeals decided that legal question because the parties had squarely presented it. 716 S.W.3d at 798-800. Both sides vigorously debated in the lower court the import of SandRidge and whether an independent contractor in Medellin's position should be able to avail himself of the necessary-use exception. See id. at 799 (“JMI, however, contends the necessary-use exception does not apply because Medellin was an independent contractor of an independent contractor.”); see also id. (noting JMI's argument that the necessary-use exception “is a narrow one ․ [;] it does not extend a duty to an employee or independent contractor because a worker can take reasonable measures to protect himself or choose not to perform unsafe aspects of the work”). The judgment necessarily rejected JMI's argument to the contrary. We may review the legal conclusions on which the lower court's judgment rests.

In concluding that we may always review questions of law actually decided below, we stand in good company. Our colleagues on the federal appellate courts have long endorsed the oft-invoked principle that they may review any issue presented to them that was pressed or passed on below. E.g., Estrada v. Smart, 107 F.4th 1254, 1263 (10th Cir. 2024) (noting that when an issue is addressed by a lower court, the higher court will “set aside [its] general rules on forfeiture and waiver”). So does the Texas Court of Criminal Appeals. See Ex parte Sanders, 663 S.W.3d 197, 202 (Tex. Crim. App. 2022). Invoking Texas Rule of Appellate Procedure 66.1—that court's analogue to our Rule 53.1—the Court of Criminal Appeals has observed that “it is not dispositive that a party may not have preserved an issue in the trial court where the court of appeals properly addressed the issue, and we granted discretionary review of it.” Id. We agree.

2

We now resolve what we left open in SandRidge. Independent contractors are an especially poor fit for the necessary-use exception. We thus hold today that an independent contractor may not avail himself of the necessary-use exception to pursue a premises-liability claim.

That holding flows from our previous decisions, and we rely on the same considerations we have invoked previously. First, the nature of the independent-contractor role presumes that before accepting a job, he is able—indeed, expected—to evaluate the requested work and make judgment calls as to the best way to mitigate dangers posed by obvious hazards throughout the course of the work. As we put it in Moritz, “one who hires an independent contractor generally expects the contractor to take into account any open and obvious premises defects in deciding how the work should be done, what equipment to use in doing it, and whether its workers need any warnings.” 257 S.W.3d at 215-16.

Indeed, independent contractors are typically better positioned to avoid premises defects. They are “hired for special projects that often entail special expertise.” Id. at 217. And much like the employee we encountered in Austin, they are frequently hired for the very purpose of fixing an unreasonably dangerous condition. It would be incongruous to permit independent contractors to assert they were unable to mitigate risk from the danger when they agreed to take the job. Cf. Austin, 465 S.W.3d at 213-14 (no premises liability for employer-landowner directing employee to clean hazard). And it would be stranger still to bar employees—who are contractually obligated to do their work—from invoking the necessary-use exception, while affording it to independent contractors who pick and choose their work. Our holding today avoids that unsound result.

Furthermore, independent contractors generally do not implicate the policy considerations that led us to recognize the necessary-use exception in Parker. There, an ordinary citizen with no special knowledge or skills was unable to safely navigate an apartment's defective exit—the only means of leaving those premises. Parker, 565 S.W.2d at 514. An independent contractor on a job site is situated differently. See Moritz, 257 S.W.3d at 217. He typically brings specialized knowledge and experience, he bears insights into the nature of the defect, he provides his own equipment, he retains authority to ensure his own safety, and he can refuse a job when it presents a danger he cannot accept. Id.; see also Austin, 465 S.W.3d at 203.

The record here bears out all those considerations. Medellin was a skilled worker with approximately three years of experience in the roofing industry. As Medellin acknowledged, he was keenly aware of the dangers inherent in working on a roof without fall protection. He further testified that he brought his own equipment to the worksite on the day of the accident. Although Medellin did not bring a harness, other workers present that day testified that they routinely bring a harness to roofing jobs. And there was evidence in the record that the ropes and lanyards associated with fall-protection systems present their own risks of tripping and falling. All of these considerations indicate that Medellin had the type of knowledge necessary to assess how best to mitigate the obvious risks associated with an unguarded roof's edge. Moritz, 257 S.W.3d at 217; see also Austin, 465 S.W.3d at 203.

To sum up, Medellin's claim sounds in premises liability. The dangerous condition that caused his injury was open and obvious. And as an independent contractor, he cannot avail himself of the necessary-use exception. Medellin's claim therefore fails as a matter of law.

V

For the foregoing reasons, we reverse the court of appeals's judgment and render a take-nothing judgment in favor of JMI.

Our premises defect and negligent activity cases define the duties of reasonable care that people and entities owe because they own, occupy, or control land. I agree with the Court that this independent contractor's negligence suit against a general contractor in control of the premises for a roofing project involves an activity that only became dangerous because of a premises condition. Accordingly, we look to our premises liability cases (not our cases addressing negligent activity on the premises) to define the duty that the general contractor owed with respect to that condition. I agree that no such duty was owed here because the dangerous condition—an unguarded roof edge—was open and obvious.

But the same person or entity can owe additional duties that spring from sources other than their occupation of land.¹ I do not understand the Court's opinion to alter the scope of the separate duty owed by one who retains control over the work of an independent contractor, which is to exercise that control with reasonable care.² Many of our cases addressing this duty have nothing to do with ownership or occupation of land.³ But in cases like Lee Lewis Construction, Inc. v. Harrison, 70 S.W.3d 778, 783 (Tex. 2001), we recognized that those who occupy or control land and also retain control over work owe this additional duty.⁴ We have never suggested that this separate duty differs when the party owing it also occupies or controls the premises, so our cases addressing the duty outside the premises context are also informative here.

As we reiterated just last year, retained control over work can support liability only if it “relate[s] directly to the injury caused by the negligence.” Massage Heights Franchising, LLC v. Hagman, 712 S.W.3d 615, 622 (Tex. 2025). I agree with the Court that in this case the record does not show that the general contractor retained control over the independent contractor's work activity that led to the injury or that the general contractor increased the risk of injury. Because the Court does not modify our retained-control precedents in reaching this conclusion but simply applies them, I join its opinion.

FOOTNOTES

1.   JMI's appeal identified several other issues we do not discuss here because they are not relevant to our decision, and we express no view on them.

2.   As noted above, we would later hold in United Scaffolding that a single workplace injury may give rise to negligent activity, or premises liability, but not both. 537 S.W.3d at 471. Moritz predates that pronouncement.

3.   To the extent Lee Lewis conflicts with our decisions in later cases, the later cases control. Moss v. Gibbs, 370 S.W.2d 452, 458 (Tex. 1963) (“[I]f there be any conflict between two cases, the later case is ordinarily held to modify the earlier holding.”); see also 16 Tex. Jur. 3d Courts § 121 (updated Mar. 2026) (“Because the Supreme Court has the power to overrule or modify its former decisions, when there is a conflict in the opinions of the Supreme Court, in accordance with the general rule, the one later in point of time will control.” (footnote omitted)).

4.   In its discussion of Medellin's negligent-activity claim, the court of appeals noted a document in evidence that referred to the presence of a watchman at the worksite. 716 S.W.3d at 797. Medellin's briefing before our court does not press any theory of liability related to a watchman. Indeed, neither the word “watchman” nor any synonym appears in his brief. In any event, there is no particularized jury finding that JMI controlled a deficient watchman.

5.   In Austin v. Kroger, we clarified that this doctrine is incorporated in the general premises-liability rule. 465 S.W.3d at 203. A landowner's duty is “to make safe or warn against any concealed, unreasonably dangerous conditions of which the landowner is, or reasonably should be aware but the invitee is not.” Id. (emphasis added). Thus, when a hazard is not concealed—that is, when it is open and obvious—the landowner owes no duty to an invitee regarding that hazard.

6.   Some doctrinal whiplash thereafter ensued, but we need not belabor it here. In short, Parker claimed to “abolish” the open-and-obvious doctrine altogether, but that did not last long, and the doctrine was swiftly revived. See generally Austin, 465 S.W.3d at 207 (discussing the post-Parker cases); see also, e.g., Nabors Drilling, U.S.A., Inc. v. Escoto, 288 S.W.3d 401, 412 (Tex. 2009); Brookshire Grocery Co. v. Goss, 262 S.W.3d 793, 794 (Tex. 2008); Moritz, 257 S.W.3d at 216; Cent. Ready Mix Concrete Co. v. Islas, 228 S.W.3d 649, 651 (Tex. 2007); and Shell Oil Co. v. Khan, 138 S.W.3d 288, 295 (Tex. 2004). The open-and-obvious doctrine remains alive and well today. See SandRidge, 642 S.W.3d at 567 (explaining that the open-and-obvious doctrine has not been abrogated).

7.   Medellin's brief uses the term “waiver,” but as we have recently noted, the alleged deficiency at issue here is actually forfeiture, not waiver. See Borusan Mannesmann Pipe US, Inc. v. Hunting Energy Servs., LLC, 716 S.W.3d 572, 574 n.1 (Tex. 2025). Forfeiture is the failure to timely assert a right or raise an issue. Bertucci v. Watkins, 709 S.W.3d 534, 541 n.5 (Tex. 2025); see also United States v. Olano, 507 U.S. 725, 733, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993). Waiver, on the other hand, speaks to “intentional relinquishment or abandonment of a known right.” Bertucci, 709 S.W.3d at 541 n.5 (quoting Olano, 507 U.S. at 733, 113 S.Ct. 1770). This opinion will use the term “forfeiture,” which captures the substance of Medellin's argument.

1.   See Austin v. Kroger Tex., L.P., 465 S.W.3d 193, 215-17 (Tex. 2015); Coastal Marine Serv. of Tex., Inc. v. Lawrence, 988 S.W.2d 223, 225-26 (Tex. 1999). In addition to the duty to reasonably exercise retained control of work, which I discuss here, other examples of separate duties that occupiers of land can owe include the duty to exercise reasonable care in undertaking to render services necessary for another's protection, e.g., Nall v. Plunkett, 404 S.W.3d 552, 555-56 (Tex. 2013); see Elephant Ins. Co. v. Kenyon, 644 S.W.3d 137, 151 (Tex. 2022), and the safety-related duties that employers owe employees, see, e.g., Austin, 465 S.W.3d at 215-17.

2.   The Restatement (Second) of Torts explains these duties and their different sources in detail. We have recognized the duty of a premises owner or occupier to carry out its activities on the premises with reasonable care, which is addressed in Sections 341 and 341A of the Restatement. See, e.g., Austin, 465 S.W.3d at 215. Our cases addressing the duty of one who retains control over the work of an independent contractor to exercise that control with reasonable care draw on Section 414 of the Restatement. See, e.g., Lee Lewis Constr., Inc. v. Harrison, 70 S.W.3d 778, 783 (Tex. 2001).

3.   E.g., Massage Heights Franchising, LLC v. Hagman, 712 S.W.3d 615, 620-23 (Tex. 2025); Gonzalez v. Ramirez, 463 S.W.3d 499, 506 (Tex. 2015); D. Houston, Inc. v. Love, 92 S.W.3d 450, 452 & n.1, 454 (Tex. 2002); Standard Ins. Co. v. McKee, 146 Tex. 183, 205 S.W.2d 362, 364-65 (1947).

4.   See also, e.g., Gen. Elec. Co. v. Moritz, 257 S.W.3d 211, 214-15 (Tex. 2008); Dow Chem. Co. v. Bright, 89 S.W.3d 602, 605-07 (Tex. 2002); Koch Refin. Co. v. Chapa, 11 S.W.3d 153, 155-56 (Tex. 1999); Clayton W. Williams, Jr., Inc. v. Olivo, 952 S.W.2d 523, 527-28 (Tex. 1997); Redinger v. Living, Inc., 689 S.W.2d 415, 418 (Tex. 1985).

Justice Hawkins delivered the opinion of the Court.

Justice Busby filed a concurring opinion.