Pilates for Knee Osteoarthritis: What the Trials Actually Show

Pilates can help people with knee osteoarthritis who are currently inactive. It has not earned a special status above conventional exercise or physiotherapy.

That distinction gets lost easily. Two 2025 meta-analyses found that Pilates improved pain and physical function compared with no intervention. In the larger question people actually face—whether to choose Pilates over another exercise program—the evidence is far less flattering. A pooled analysis found no statistically significant pain advantage over conventional exercise, and the only direct trial against suspension training favored suspension training on every measured outcome. Pilates is a reasonable way to move an arthritic knee. It is not a proven superior way.

Versus inactivity, Pilates has a meaningful signal

A 2025 systematic review and meta-analysis by Oliveira et al. pooled randomized trials of Pilates for knee osteoarthritis. Against no intervention, Pilates reduced pain with a standardized mean difference (SMD) of −1.09. The review also found statistically significant improvements in physical function and quality of life versus no intervention.

An SMD expresses the difference between groups in standard-deviation units rather than in a familiar pain scale such as 0 to 10. The result points toward less pain with Pilates, but it cannot tell an individual patient exactly how many points their pain score will change. The review did not provide a confidence interval for that pooled estimate in the evidence summarized here.

An independent 2025 meta-analysis by Zhang et al. reached the same broad conclusion: Pilates improved pain and function compared with no intervention and appeared safe in the included randomized trials. That replication matters. It supports the practical conclusion that a person with knee osteoarthritis who is doing little or no exercise can reasonably start Pilates.

Neither review proves that Pilates changes the course of osteoarthritis or repairs cartilage. These were short exercise interventions, typically lasting weeks, and neither review identified follow-up beyond the active program. The studies measured symptoms and function during treatment, not whether benefits persisted six months later.

The advantage disappears against active exercise

The Oliveira review found no statistically significant superiority of Pilates over conventional exercise for pain. The confidence intervals overlapped in the active-comparator analyses. Pilates did show an advantage for knee range of motion, but that is one endpoint, not proof of better overall symptom control or long-term joint health.

This is the central finding for someone choosing between a Pilates class and a physiotherapist-guided strengthening program. Both may be sensible. Current evidence does not show that Pilates relieves knee pain better.

That fits clinical guidance. The American College of Rheumatology, Osteoarthritis Research Society International, and American Academy of Orthopaedic Surgeons endorse exercise as first-line care for knee osteoarthritis without naming Pilates as the preferred modality, as summarized by Zhang et al.. A 2024 systematic review of physiotherapy techniques also describes an established evidence base for approaches such as quadriceps strengthening, manual therapy, neuromuscular electrical stimulation, therapeutic ultrasound, and combined protocols.

Other structured options have evidence, too. A 2026 randomized trial of online, unsupervised Tai Chi found improved knee pain and function versus control. A 2024 meta-analysis in postmenopausal women with knee osteoarthritis found clinically meaningful benefits of physical exercise across modalities for muscle function, pain, and quality of life. The point is not that every program performs identically. It is that the evidence does not support treating Pilates as the lone answer.

One small trial favored suspension training

The only direct randomized comparison of Pilates and suspension training came from Karimi et al. in 2021. The trial enrolled 30 women with knee osteoarthritis, with a mean age of 61.4 years; 10 participants entered each of three groups: Pilates, TRX-style suspension training, or passive control. The active programs ran for eight weeks, three times per week, for 60 minutes per session.

Both active groups improved versus passive control on dynamic balance measured by the Y-balance test, static balance measured by the Stork stand, knee range of motion measured by goniometry, and WOMAC motor function. Suspension training was statistically superior to Pilates on all four outcomes (P ≤ 0.05). No outcome favored Pilates.

That pattern is interesting, but the trial cannot settle the question. Ten people per group is too few for confident between-group conclusions, the study enrolled women only, and the authors did not report between-group mean differences, standardized effects, or 95% confidence intervals. Without those estimates, readers cannot judge the size or precision of the apparent advantage. Suspension training may impose greater balance and stabilization demands than mat-based Pilates, but that explanation remains plausible inference rather than a tested mechanism in this trial.

Pilates beats inactivity in the available trials. It has not beaten conventional exercise on pain, and one underpowered direct comparison favored suspension training.

Why the pooled number needs restraint

The SMD of −1.09 sounds decisive, but Oliveira et al. rated the certainty of evidence as low for every Pilates outcome. The review downgraded confidence for risk of bias, imprecision, and substantial inconsistency across trials. Heterogeneity ranged from I² = 66% to 91%, meaning results varied considerably among studies.

Those trials did not all test the same intervention. Programs differed in whether they used mat work or apparatus, ran two or three times weekly, lasted 45 to 60 minutes per session, and took place under clinic supervision or at home. Pooling them produces a useful directional estimate, but not a prescription for the ideal Pilates dose or a guarantee that every version works equally well.

Exercise trials also face an unavoidable problem: participants know whether they are attending Pilates. Most included trials raised at least some concerns on Cochrane's risk-of-bias tool, especially around allocation concealment and assessor blinding. Saleem et al.'s 2022 trial of supervised mat-based Pilates in women reported double-blinding and found improvements in pain, range of motion, and functional disability versus control. Still, participants cannot truly be blinded to the exercise they perform. That limitation applies across exercise research.

The evidence also leaves obvious clinical questions unanswered. Neither 2025 meta-analysis reported results by radiographic disease severity, body mass index, age, or baseline pain. No included trial tracked whether improvements survived after classes ended. The studies do not establish which patients respond best, whether men respond similarly, or whether Pilates produces lasting benefits.

The practical choice is often the program you will keep doing

Zhang et al. noted that about 87% of people with knee osteoarthritis fail to meet World Health Organization physical-activity targets. That is a larger real-world problem than the difference between two exercise labels. A class that gives someone a schedule, coaching, and a reason to show up may outperform an excellent home program that remains a PDF on a phone.

There is no direct evidence here that Pilates produces better adherence than conventional physiotherapy. Its structured format may help some people stay engaged; for others, cost, travel, floor transfers, or an intimidating studio may do the opposite. Home-based exercise can also reduce barriers, although adherence varies with support and circumstance. A 2025 narrative review of home-based exercise describes both its potential benefits and the practical barriers that commonly affect participation.

If Pilates is accessible, tolerable, and enjoyable enough that you will attend consistently, it is a defensible choice for knee osteoarthritis. If a clinician has prescribed progressive strengthening, balance work, or conventional physiotherapy, current evidence gives no reason to replace it solely because Pilates carries a large pooled SMD against inactivity. If suspension training interests you, the one small trial is a reason to discuss a properly supervised program—not a reason to declare a winner.

Keep medication and clinician-supervised care in the plan when they are indicated. Exercise is first-line treatment for knee osteoarthritis, but Pilates is one route into that treatment, not a demonstrated upgrade over every other route.