Design: Cochrane systematic review and random-effects meta-analysis, GRADE assessed
Population: 2,672 participants across 53 randomised trials, mean age 61.9, mostly ambulatory subacute or chronic stroke
Primary outcome: death, disability, adverse events, risk factors, fitness, walking and physical function
Effect: mortality RD 0.00 (−0.01 to 0.01, high certainty); secondary events RD −0.00 (−0.03 to 0.02, high certainty); VO2 +2.37 (1.39–3.36) against MCID +3.5

This Cochrane review pooled 53 randomised trials and 2,672 participants, average age 61.9 years, comparing cardiorespiratory training after stroke with usual care, no intervention or a non-exercise control. Most participants were ambulatory and in the early subacute or chronic phase.

The negative findings carry the strongest evidence. Training does not change mortality at the end of the intervention (risk difference 0.00, 95% CI −0.01 to 0.01; 36 studies, high certainty) or at the end of follow-up (high certainty), and does not change the incidence of secondary cardiovascular or cerebrovascular events at the end of intervention (RD −0.00, −0.03 to 0.02; high certainty), probably not at follow-up either. That matters because secondary prevention is the reason exercise is often prescribed.

What it may do is smaller than usually implied. Cardiorespiratory fitness probably improved slightly (VO2 +2.37 ml/kg/min, 1.39 to 3.36; moderate certainty) against a minimal clinically important difference of +3.5. Comfortable walking speed probably increased slightly (+0.08 m/s, 0.04 to 0.12; moderate certainty) against an MCID of +0.13 — so the gain is real and, on these figures, below the threshold that patients notice. Disability and balance may improve, but the evidence is graded very uncertain and the point estimates sit near their MCIDs. Longer interventions did better in subgroup analysis.

None of this is a reason to stop exercise after stroke, which is safe and recommended. It is a reason to describe it honestly: for function and fitness, not as a way of preventing the next stroke.

Prescribe exercise after stroke for function and fitness, not as secondary prevention
Keep blood pressure, lipid and antithrombotic management as the secondary prevention plan
Favour programmes of 12 weeks or more — subgroup analysis favoured longer interventions
Reassure patients about safety: mortality and secondary events were unchanged, with high-certainty evidence

Why it matters

The rationale most often given to patients for post-stroke exercise is the one the evidence does not support.

The statistics, in plain English

High certainty on a null result is a strong statement: it means further trials are unlikely to reveal a mortality or recurrence benefit that was missed. For the positive outcomes, the pattern is consistent — the confidence intervals exclude zero but sit below or around the minimal clinically important difference, meaning the effect is statistically real and probably too small for a patient to feel. Very low certainty on disability and balance reflects imbalanced exercise exposure between groups in 28 of the 53 studies.