Modest amounts of exercise eased chronic back pain in 239 trials, with little extra from more

Exercise is one of the most common treatments for back pain that will not go away, and guidelines put it near the top of the list. What has been less clear is how much of it people need: how often, for how long and over how many weeks.

A review in the British Journal of Sports Medicine pooled 239 randomized trials to model the dose. Low to moderate amounts of exercise were linked to short-term improvements in pain and disability, and doing more added little. The evidence behind those estimates was weak, and the authors caution against treating any single dose as the answer.

Why exercise is a standard treatment for back pain

Low back pain has been the leading cause of disability globally for at least the past three decades, according to the Cochrane review of exercise therapy. Most episodes settle within weeks. Pain that lasts longer is called chronic; the Cochrane review uses the definition of chronic non-specific low back pain of more than 12 weeks’ duration, and that longer-lasting kind is what the new review studied.

The case that exercise helps chronic back pain in general comes from earlier reviews. A 2021 Cochrane review of 249 trials found moderate-certainty evidence that exercise treatment is more effective for pain than no treatment, usual care or placebo. The gain for everyday function was smaller, and fell short of the threshold its authors had set for a meaningful difference.

Guidelines reflect that evidence. The American College of Physicians’ 2017 guideline, published in the Annals of Internal Medicine, says that for chronic low back pain, clinicians and patients should initially select nonpharmacologic treatment with exercise, or with other non-drug options such as tai chi, yoga and mindfulness-based stress reduction, before turning to medicines.

How the back pain exercise review worked

The researchers searched six databases, two trial registries and citation lists up to June 2025. They included randomized trials of exercise in adults with chronic low back pain that measured pain or disability. Almost all of the 239 trials studied non-specific pain, meaning back pain without an identified cause.

Rather than asking only whether exercise worked, the team used a dose-response network meta-analysis. That is a method that combines trials of different exercise amounts to estimate how results change as the dose goes up. Dose meant how many sessions a week, how long each session lasted and how many weeks the program ran.

Pain and disability were scored from 0 to 100. A change of 5 points or more was set in advance as clinically meaningful, meaning large enough to matter to patients. Short term meant up to three months after starting, with separate estimates for later time points.

What exercise dose eased back pain and disability

For short-term disability, the model-predicted best dose was four sessions a week of about 25 minutes, over 13 weeks. For short-term pain, it was three sessions a week of about 30 minutes, over 8 weeks. Neither is a large commitment compared with the amounts some exercise programs ask for.

Those are best estimates. The authors note that credible intervals of most dose-levels overlapped, meaning the ranges of likely results for different doses were too similar to separate them with confidence. Overall, exercise at low to moderate doses was linked to improvements, with limited additional benefit at higher exposure.

The gains also faded with time. Few dose combinations were clinically meaningful beyond the short term, so the review says little about whether any amount of exercise keeps pain down over a year or more. That matches a common pattern in back pain research, where early improvements are easier to show than lasting ones.

Where the back pain dose findings fit

The same research network has looked at which kinds of exercise help most. A 2019 network meta-analysis in the British Journal of Sports Medicine pooled 89 trials with 5,578 patients. It found low quality evidence that Pilates, stabilisation/motor control, resistance training and aerobic exercise training are the most effective treatments, depending on whether pain, function or mental health was measured. Stretching and McKenzie exercises, a method built on repeated back movements, did not clearly beat no treatment.

An independent review published in July 2026 in the Journal of Orthopaedic and Sports Physical Therapy asked a similar dose question and reached a compatible answer. Across 70 randomized trials with 3,991 people, it found that higher-frequency exercise programs were associated with improvements in disability, with three or more sessions a week doing better over the long term than fewer. Yoga, balance-based and assisted exercise programs offered modest extra benefits. Its authors, too, stressed that the certainty was low and that the gains fell short of what they had set as clinically meaningful, and concluded that people’s preferences, what is feasible and whether they keep going matter more than any single element of the dose.

How much to trust the back pain findings

The quality of the underlying trials is the main limit. The authors judged that 84% of trials were at high risk of bias, and rated the certainty of the evidence as very low to low. Many exercise trials are small, and people cannot be blinded to whether they are exercising, which can inflate reported benefits for outcomes like pain.

Harms looked minor.

Adverse events were reported in 28.5% of trials and were mostly non-serious muscle and joint problems. The earlier Cochrane review similarly found mostly minor harms, e.g. muscle soreness, in people doing exercise treatment.

Some practical questions went unanswered. Only two trials included people whose pain spreads from the back into the leg, known as radicular pain. The authors describe exercise for those presentations as a key evidence gap.

What the back pain review cannot tell us

The model estimates doses by comparing across trials that used different people, programs and kinds of exercise. It cannot say which type of exercise works best for a given person, or whether the estimated doses would hold in any one program. “Optimal” here means the dose the statistical model predicts, not a dose proven better than others in a head-to-head trial. Most trials followed people for weeks or months, not years. Because most enrolled people with non-specific pain, the results say little about back pain with a known cause, such as a fracture or a trapped nerve.

What this adds to exercise for back pain

The review supports what guidelines already say, that exercise helps chronic low back pain in the short term. It adds an estimate that modest amounts capture most of that benefit. The authors conclude that the available evidence does not support a single optimal exercise dose, and that the estimates should be read with caution given the weak evidence.

This does not change current guidance. With certainty rated low to very low, the estimated doses are a starting point for better trials. They are not a settled answer on how much exercise is enough.