Objective: To systematically review the efficacy of exercise interventions on pain relief in patients with fibromyalgia (FM). Methods: We performed a comprehensive search of 8 western databases and reference lists through March 2009. We included randomized controlled trials (RCTs) with pain as an endpoint, measured by the Fibromyalgia Impact Questionnaire (FIQ) or a visual analogue scale (VAS). The exercise treatments were compared with a no-exercise control group. Study quality was assessed by the Jadad scale. We calculated effect sizes to assess the magnitude of treatment effect, and meta-analyzed for overall effect. Results: Forty-five studies were critically appraised for effects on pain. Ten RCTs published between 1992 and 2008, conducted in Europe and North America met eligibility criteria with 767 participants (98% women). Mean age was 47 years and mean symptom duration was 10 years. The meta-analysis results showed that 6 to 24 weeks of aerobic, strength training, pool and multi-component exercise had a statistically significant decrease in pain outcomes versus education, usual care and non-exercise controls. The pooled effect size was 0.45 (95% Confidence Interval, 0.09 to 0.80). Conclusion: Physical exercise may have positive effects on pain reduction in patients with FM. Long-term, rigorous and well-controlled randomized trials are warranted.