Are Leg Extensions Bad for Your Knees? What the Research Actually Says
If you've ever been told that leg extensions are “bad for your knees,” you are very much not alone. The leg extension machine has somehow become one of the villains of the fitness and rehab world, right alongside letting your knees travel over your toes, squatting below parallel and probably whatever other completely normal human movement we're collectively afraid of this week. If you've previously injured your ACL or had an ACL reconstruction, there's a good chance you've heard even stronger warnings about avoiding them because they supposedly place too much strain on the ACL.
I think leg extensions have gotten a much worse reputation than they deserve. Contrary to popular belief, they're not inherently bad for your knees, including for someone who has injured their ACL or is going through ACL reconstruction rehab. We actually have a decent amount of research supporting open-chain knee extension, meaning the foot isn't planted on the ground, as part of a comprehensive ACL rehab program. I'd even argue that an ACL rehab program is probably lacking if it never includes some form of direct knee extension strengthening, because restoring quadriceps strength is such an important part of getting that knee back to doing all the things we eventually expect it to do.
Obviously, that doesn't mean I'd take someone who is three weeks out from ACL reconstruction, slap the whole stack on the leg extension machine and tell them to send it. How much resistance we use, what range of motion we train through and when we introduce or progress an exercise all matter, particularly when we're dealing with a healing graft. Unfortunately, a lot of conversations about knee rehab completely bulldoze over that nuance and divide exercises into neat little categories of “good for your knees” and “bad for your knees,” when that's rarely how rehabilitation actually works.
The bigger conversation here isn't really about whether one machine at the gym is good or bad. It's about understanding the difference between loading a knee and “damaging” a knee, because those two things have gotten mashed together in a lot of fitness and rehab advice. It's also about understanding what an MRI can and cannot tell us, why structural findings like arthritis and meniscus tears don't automatically explain someone's pain, and even why some of the assumptions we've historically made about ACL injuries themselves—including the idea that a torn ACL simply cannot heal without surgery—are becoming a whole lot more complicated as new research emerges.
So, yes, we're going to talk about leg extensions. But we're also going to talk about why having an injured knee, a reconstructed ACL or an ugly-looking MRI does not automatically mean you have a fragile knee that needs to spend the rest of its life being protected from load.
Why did leg extensions get such a bad reputation?
The concern around leg extensions isn't completely fabricated, which is important to acknowledge before we go swinging too far in the opposite direction. A leg extension is considered an open kinetic chain exercise because the foot is free to move, whereas something like a squat is considered closed-chain because the feet remain planted while the body moves around them. Historically, ACL rehabilitation tended to favor closed-chain exercises and approach open-chain knee extension much more cautiously because of what happens biomechanically at the knee during the movement.
When your quadriceps contract during a leg extension, they can create anterior translation of the tibia relative to the femur, particularly as the knee approaches full extension. Because one of the ACL's jobs is to resist excessive anterior translation of the tibia, the ligament experiences strain during portions of the movement. That is a real biomechanical phenomenon, and it's one of the reasons I may be more thoughtful about range of motion and resistance when introducing knee extensions early after ACL reconstruction.
Where we went off the rails was taking the statement “this movement places strain on the ACL” and turning it into “this movement is bad for the ACL.” Those aren't interchangeable statements. Your ACL experiences force during normal human movement, and eventually someone recovering from an ACL injury may need that knee to tolerate sprinting, jumping, landing, cutting, decelerating and lifting heavy shit. If our entire rehabilitation philosophy revolves around preventing the ACL from ever experiencing meaningful stress, we've got a pretty enormous problem when that person eventually needs to go back to actually using their knee.
This is where progressive loading becomes really important. Healing and injured tissues are loaded all the time in physical therapy because applying an appropriate amount of stress is part of how we restore their capacity. The goal isn't to eliminate load; it's to manipulate it. We can change resistance, range of motion, volume, frequency, tempo and dozens of other variables based on where someone is in the healing process and how their knee is responding. There's a massive difference between gradually exposing a healing knee to an appropriate stimulus and asking that knee to tolerate more than it's currently prepared for, and reducing both situations to “loading the ACL” isn't particularly useful.
What does the research actually say about leg extensions and ACL strain?
One of the classic studies in this area actually measured ACL strain while people performed both squatting and active knee flexion and extension. Interestingly, the maximum ACL strain measured during the squat was not significantly different from the maximum strain measured during the open-chain movement (PMID: 9397272). That doesn't mean every squat and every leg extension place identical forces on the ACL throughout every degree of knee motion, because they don't. What it does tell us is that the old-school idea that closed-chain exercise is automatically safe while open-chain exercise is automatically dangerous is far too simplistic.
That's an important distinction because people LOVE exercise rules. “Never let your knees go over your toes” is easy to remember. “Never do leg extensions after an ACL injury” is easy to remember. Understanding that load changes depending on joint angle, resistance, healing stage, graft type, exercise selection and the individual standing in front of you requires considerably more thought, which is unfortunately much harder to turn into a sexy Instagram infographic.
The clinical research gives us even more reason to move away from blanket avoidance. Fukuda and colleagues conducted a randomized controlled trial comparing earlier versus later introduction of open-chain knee extension following ACL reconstruction using a hamstring autograft. The earlier group began knee extensions four weeks after surgery through a restricted range of approximately 90 to 45 degrees of knee flexion, while the later group waited until 12 weeks to introduce them through a larger range. Importantly, beginning knee extensions earlier did not result in greater anterior knee laxity, and the earlier group actually recovered quadriceps strength faster (PMID: 23423316).
That is a much more useful way to think about this exercise clinically. The question isn't, “Can someone with an ACL reconstruction do leg extensions: yes or no?” I want to know how far out they are from surgery, what graft they had, how the knee is behaving, what range we're working through, how much resistance we're using and what we're trying to accomplish. Someone four weeks post-op performing controlled knee extensions through a deliberately restricted range is a completely different scenario from someone four weeks post-op grinding out maximal full-range sets, even though both could technically be described as “doing leg extensions after ACL surgery.”
We also have more than individual trials to work with. A 2024 systematic review and meta-analysis included nine randomized trials comparing open- and closed-chain exercises during ACL rehabilitation. Open-chain exercise showed an advantage for quadriceps strength at certain time points around three to four months, without evidence that it produced worse knee laxity. The authors concluded that open-chain exercise has an important place in ACL rehabilitation, although the certainty of the available evidence was rated low to very low, which is worth acknowledging rather than pretending science handed us an unquestionable commandment carved into a stone tablet (PMID: 38887689).
A 2025 systematic review and meta-analysis looked specifically at open-chain exercise during the earlier stages after ACL reconstruction and reached a similar overall conclusion. Protocols and study quality varied, so we still don't have one magical week where every person on Earth should start hammering leg extensions. Overall, however, open-chain exercise was associated with improvements in strength, function, patient-reported outcomes and return to play, particularly when introduced at least four weeks after surgery, without evidence of negative effects from appropriately programmed open-chain exercise during those earlier stages (PMID: 39985872).
The quad-strength problem we aren't talking about enough
When the entire ACL conversation revolves around protecting the graft, we can lose sight of one of the major problems we're actually trying to solve: the quad gets weak as hell.
Significant quadriceps weakness is incredibly common following ACL injury and reconstruction, and it can persist for a LONG time when it isn't adequately addressed. That's not a deficit I want to shrug off, because your quads have a pretty important job. They contribute to producing and absorbing force at the knee during walking, running, climbing stairs, squatting, jumping, landing and decelerating, which happens to be a decent chunk of the stuff we eventually want people to get back to doing.
Squats, lunges, split squats, step-ups and other weight-bearing exercises are fantastic, and I absolutely want them in a comprehensive ACL rehab program. The problem is that humans are also spectacular compensators. Someone with a weak surgical-side quad can perform something that looks remarkably squat-shaped while subtly shifting toward the uninvolved leg, changing what they're doing at the hip, altering pressure through their foot or finding another strategy that allows them to complete the task without loading the involved quad as much as we think they are.
Put someone on a knee-extension machine and suddenly there's nowhere for that quad weakness to hide. I can directly load the muscle, compare sides, progressively increase resistance and actually make the weak thing stronger without allowing the rest of the body nearly as many opportunities to swoop in and save it. That's not an argument for replacing squats and split squats with leg extensions; it's an argument for recognizing that both have a place, and pretending compound exercises automatically take care of every strength deficit can leave some pretty significant holes in an ACL rehab program.
So, how should leg extensions actually be used during ACL rehab?
This is where individualized rehabilitation matters. There are post-operative protocols and general tissue-healing timelines that help guide decision-making, but a protocol isn't supposed to replace clinical reasoning or turn every ACL patient into the exact same knee attached to a different human.
Depending on the surgical procedure, graft choice, stage of healing, symptoms, swelling, range of motion, strength and goals, I can manipulate the range of motion, resistance, sets and reps, tempo, isometric versus dynamic contractions, proximity to failure, weekly training volume and frequency. Earlier in rehab, that may mean using lighter resistance through a more restricted range. As healing progresses and the knee demonstrates greater capacity, I can gradually expose it to larger ranges and heavier loads.
Eventually, however, I'm not particularly interested in keeping someone permanently trapped inside a “safe” rehab range. Life doesn't happen exclusively between 90 and 45 degrees of knee flexion, and neither does sport. If someone ultimately needs to tolerate deep knee flexion, terminal extension, sprinting, landing, decelerating or heavy lifting, rehabilitation has to progressively prepare them for those positions and forces rather than endlessly avoiding them.
And while we're dismantling outdated ACL rules, there's another one worth discussing because the research around it has gotten very interesting over the last several years.
Can a torn ACL actually heal without surgery?
For a long time, patients with complete ACL ruptures were often told some version of, “The ACL can't heal itself.” That statement is increasingly difficult to defend as an absolute because we now have MRI evidence showing continuity of previously ruptured ACLs in some people who were managed without reconstruction.
One of the more interesting pieces of evidence comes from a secondary analysis of the KANON trial, which followed adults with acute ACL ruptures who were randomized to early ACL reconstruction or rehabilitation with the option of delayed reconstruction. When researchers went back and evaluated MRI findings among participants initially managed with rehabilitation, approximately one-third of that group demonstrated evidence of ACL healing at two years. Among participants who remained managed with rehabilitation alone and never crossed over to reconstruction, approximately half demonstrated evidence of healing. Even more interestingly, participants whose ACLs showed evidence of healing reported favorable sport/recreation function and quality-of-life outcomes at two years (PMID: 36328403).
There are about seventeen caveats I want to staple to that paragraph before someone interprets it as “ACL surgery is pointless.” This does not mean every ACL rupture will heal, nor does it tell us that every person with an ACL injury should choose non-operative management. Surgical decision-making can depend on the type and location of the tear, associated injuries, instability, age, activity demands, sport, personal goals and a bunch of other individual factors. What the research does tell us is that saying a ruptured ACL categorically cannot heal is no longer consistent with the evidence we have.
The Cross Bracing Protocol has added another fascinating layer to this conversation. In a 2023 study, 80 people with acute ACL ruptures underwent a specific non-operative protocol in which the knee was initially immobilized at approximately 90 degrees of flexion for four weeks, followed by progressive increases in range of motion and supervised rehabilitation. At three months, 90% demonstrated evidence of ACL continuity on MRI, and participants with greater evidence of healing tended to have better knee laxity, patient-reported outcomes and return-to-sport outcomes (PMID: 37316199).
Those numbers are impressive, but this is exactly where social media can take an interesting study and run screaming several miles beyond what it actually proved. The study did not have a randomized control group, there were reinjuries, and we still need higher-quality comparative research to determine which patients are appropriate candidates, whether the bracing protocol itself facilitates healing, how outcomes compare with other non-operative approaches and reconstruction, and what all of this looks like over the long term.
Longer-term evidence from KANON gives us another reason to keep our science pants on. An 11-year follow-up analysis found that ACL continuity seen on five-year MRI was not associated with superior long-term patient-reported outcomes and was actually associated with worse outcomes than reconstructed groups in some comparisons (PMID: 40387842). In other words, an ACL looking continuous again on an MRI and a person having a fantastic knee for the next decade are not necessarily the same outcome.
The most accurate thing we can currently say is that some ruptured ACLs appear capable of regaining continuity without reconstruction, and we're still figuring out who is most likely to heal, whether particular rehabilitation strategies can facilitate that process and what MRI-defined healing ultimately means for long-term function. That's considerably less exciting than declaring that everyone needs surgery OR nobody needs surgery, but unfortunately nuance continues to ruin perfectly good internet arguments.
Your MRI is not a pain-o-meter
This brings me to an even bigger knee conversation, because the way we interpret imaging can have a massive influence on how people think about their bodies.
The usual sequence goes something like this: your knee hurts, so you get an MRI. The MRI finds a meniscus tear, cartilage changes, osteophytes, arthritis or some other structural abnormality. You understandably assume that you've found the exact reason your knee hurts, and suddenly every squat, staircase or leg extension feels like it might be grinding that structural abnormality into oblivion.
Sometimes an imaging finding absolutely is relevant to someone's symptoms. I'm not arguing that MRIs are useless or that structure never matters. What I am saying is that imaging findings are incredibly common in people who don't have pain at all, which means we cannot automatically treat every abnormality on a scan as the smoking gun.
A systematic review and meta-analysis by Culvenor and colleagues looked at MRI findings across 63 studies involving 4,751 asymptomatic adults and 5,397 knees. These were people who weren't presenting because their knees hurt, yet cartilage defects were present in approximately 24% of knees, meniscal tears in about 10%, bone marrow lesions in about 18% and osteophytes in about 25%. Those findings became considerably more common with age; in groups with a mean age of 40 or older, approximately 43% had cartilage defects and 19% had meniscal tears despite being asymptomatic (PMID: 29886437).
Another study performed MRI scans on 230 knees belonging to 115 asymptomatic adults and found abnormalities in an absolutely wild 97% of knees. Meniscal tears were present in 30%, and more than half had patellofemoral cartilage abnormalities despite having no knee symptoms (PMID: 32060622).
That doesn't mean a meniscus tear can never hurt or osteoarthritis is fake. Of course those conditions can be symptomatic, and certain structural findings may be clinically important. It means that the presence of structural change does not guarantee pain, and the severity of an MRI report doesn't perfectly dictate how someone feels, functions or what their knee is capable of doing.
This distinction becomes especially important as we get older because structural changes become increasingly common. If we scanned enough pain-free 40-, 50- and 60-year-old knees, we'd find plenty of things that sound downright horrifying in radiology language attached to people who are happily walking, running, squatting and going about their lives without knowing those findings exist.
Meniscus tears, arthritis and patellofemoral pain deserve more context
Meniscus tears are probably one of the easiest examples of why context matters. An acute traumatic meniscus injury can absolutely cause symptoms, particularly when the history, mechanism and clinical presentation line up with the imaging. But meniscal abnormalities are also found regularly in people without knee pain, particularly as they get older. In that large systematic review of asymptomatic knees, meniscal tears were found in approximately 4% of younger groups and 19% of groups with a mean age of 40 or older (PMID: 29886437).
So if you're 45, develop knee pain and your MRI says you have a meniscus tear, I don't automatically get to point at the scan and say, “AHA! THERE IT IS.” Maybe that tear is clinically relevant. Maybe it has been quietly hanging out in your knee for years while you hiked, ran, squatted and lived your life without having the slightest clue it existed. I still need to know how your symptoms started, where they are, what aggravates them, what your knee can currently tolerate, how strong you are and what your physical examination looks like before deciding what role that MRI finding plays in the overall picture.
Osteoarthritis deserves the same nuance because words like “degeneration,” “cartilage loss,” “bone-on-bone” and “wear and tear” can make people feel like their knee is a mechanical object slowly disintegrating every time they use it. Osteoarthritis can absolutely be painful and disabling, but structural severity and symptoms do not have a clean one-to-one relationship. Cartilage changes, osteophytes and other features associated with OA can be found in people without knee pain, which is why an image alone cannot tell me how much someone should hurt or what they're capable of doing.
Patellofemoral pain gives us an even better example because it's particularly relevant to the leg extension conversation. People with pain around or behind the kneecap are frequently told to avoid knee-extension exercises because the assumption is that something structurally wrong with the patellofemoral joint must be responsible for the pain. A study comparing 64 people with patellofemoral pain with 70 healthy controls found minor patellar cartilage defects in 23% of the pain group and 21% of the healthy group. Patellar bone marrow lesions were present in 53% versus 51%, and high signal in Hoffa's fat pad was present in 58% versus 51%. After accounting for factors including age, BMI, sex and sports participation, none of the structural abnormalities evaluated were significantly associated with patellofemoral pain (PMID: 27206691).
That's a pretty spectacular reminder that we treat people, not pictures. If two people can have remarkably similar structural findings and one has pain while the other doesn't, there is clearly more contributing to the pain experience than the presence or absence of one abnormal-looking structure on an MRI.
So what if leg extensions actually hurt?
After all of that, we can finally come back around to the supposedly terrifying machine that started this entire conversation. If someone tells me leg extensions hurt their knees, my response isn't to tell them they're imagining it, and it also isn't to immediately conclude that the exercise is damaging their joint. I want more information because pain during an exercise gives me something to work with.
I want to know where the pain occurs, when in the range it starts, how much resistance they're using, how many sets they're doing, how frequently they're training their quads, what else is in their program and what happens to their symptoms afterward. If I reduce the resistance, does it change? What if I modify the range? What if we slow down the movement, reduce the volume or temporarily change how frequently they're doing it? Those questions are a lot more useful than simply adding “leg extensions” to an ever-growing list of forbidden exercises.
Someone with patellofemoral pain, osteoarthritis, a previous meniscus injury or another knee condition may still do perfectly well with leg extensions. We may need to adjust the range of motion, resistance, tempo or total volume based on their current capacity, but that's very different from declaring the exercise harmful.
I'm also perfectly comfortable with some discomfort during exercise for many rehab clients. A general guideline I often use is keeping symptoms around 3/10 or less, making sure pain isn't progressively ramping up throughout the activity and paying attention to how the person feels afterward. That isn't a universal law and there are situations where pain warrants a different response, but the presence of mild discomfort by itself does not tell us that tissue damage is occurring.
Pain is information. It's worth listening to, but we don't have to treat every whisper from the knee like an emergency evacuation alarm.
Maybe you don't have “bad knees” after all
I know calling something a “bad knee” sounds harmless, but I genuinely think the language we use around injuries matters because it shapes what people believe their bodies can tolerate. If you've been repeatedly told that your knee is damaged, degenerative, unstable or worn out, it's pretty understandable that you might start treating movement and loading as things that need to be approached with fear.
A history of an ACL tear doesn't automatically make your knee fragile. A meniscus tear on an MRI doesn't automatically explain your pain. Cartilage changes don't mean every squat is wearing your joint away, and having pain during a leg extension doesn't automatically mean the exercise is damaging you. You may have a knee that's been injured, operated on, irritated, deconditioned or under-loaded for a long time, and those things absolutely deserve thoughtful rehabilitation. They're also much more useful descriptions than simply deciding you own a “bad knee” now.
Ultimately, ACL rehab and knee rehab in general have to prepare people for the things they actually want to do. If your goal is to run, jump, hike, squat, lift heavy, play soccer, chase your kids around or simply walk up a flight of stairs without thinking about your knee, then your knee needs enough capacity to tolerate the forces associated with those activities.
My job as a physical therapist isn't to protect your knee from meaningful load forever. If I do that, I'm failing you just as much as I would be if I loaded you to the absolute max on day one of rehab. Instead, good rehabilitation progressively closes the gap between what your knee can tolerate today and what you eventually need it to tolerate, using symptoms, strength, function, tissue healing and your individual goals to determine how quickly we move along that continuum.
Sometimes that means squatting. Sometimes it means running. Sometimes it means jumping, landing or lifting something really freaking heavy.
And, yes, sometimes it means sitting down on the supposedly terrifying leg extension machine and getting your quads strong as hell.
Research referenced
Beynnon et al. — Compared ACL strain during squatting and active open-chain knee flexion/extension. Maximum ACL strain was not significantly different between the two exercises.
PMID: 9397272
Fukuda et al. — Randomized controlled trial comparing earlier restricted-range open-chain knee extension beginning four weeks after ACL reconstruction with later introduction. Earlier introduction did not increase anterior knee laxity and resulted in faster quadriceps strength recovery.
PMID: 23423316
Pamboris et al. — 2024 systematic review and meta-analysis of nine randomized trials comparing open- and closed-chain exercise during ACL rehabilitation. Open-chain exercise showed advantages for quadriceps strength at some time points without evidence of worse knee laxity, although certainty of evidence was low to very low.
PMID: 38887689
Fontanier et al. — 2025 systematic review and meta-analysis examining open-chain exercise during the early stages following ACL reconstruction. Open-chain exercise was associated with improvements in strength, function and other clinical outcomes without evidence of adverse effects when appropriately introduced.
PMID: 39985872
Filbay et al. — Secondary analysis of the KANON trial examining MRI evidence of ACL healing in people initially managed with rehabilitation and optional delayed reconstruction. A substantial proportion demonstrated ACL continuity on MRI, and those showing healing reported favorable two-year outcomes.
PMID: 36328403
Filbay et al., Cross Bracing Protocol — Study of 80 people with acute ACL rupture treated using a specific bracing and rehabilitation protocol. At three months, 90% demonstrated MRI evidence of ACL continuity. The study was observational and lacked a randomized control group, so these findings should not be interpreted as evidence that all ACL tears can or should be managed using this protocol.
PMID: 37316199
KANON long-term follow-up — Examined the relationship between ACL continuity on five-year MRI and outcomes at approximately 11 years. MRI continuity was not associated with superior long-term patient-reported outcomes and was associated with worse outcomes than reconstructed groups in some comparisons.
PMID: 40387842
Culvenor et al. — Systematic review and meta-analysis of 63 studies representing 5,397 asymptomatic knees. Cartilage defects, meniscal tears, bone marrow lesions and osteophytes were all commonly found in people without knee pain, with prevalence generally increasing with age.
PMID: 29886437
Horga et al. — MRI study of 230 asymptomatic knees found abnormalities in 97%, including meniscal tears in 30% and frequent cartilage and bone marrow abnormalities despite the participants having no knee symptoms.
PMID: 32060622
van der Heijden et al. — Compared MRI findings in people with patellofemoral pain and asymptomatic controls. Several structural abnormalities occurred at similar rates in both groups, and the evaluated structural findings were not significantly associated with patellofemoral pain.
PMID: 27206691
