A previous systematic review reported that topical NSAIDs were effective in relieving pain in acute conditions like sprains and strains, with differences between individual drugs for efficacy. More trials, a better understanding of trial quality and bias, and a reclassification of certain drugs necessitate a new review.Studies were identified by searching electronic databases and writing to manufacturers. We selected randomised double blind trials comparing topical NSAID with either placebo or another active treatment in adults with acute pain, and extracted dichotomous information approximating to a 50% reduction in pain at one week, together with details of adverse events and withdrawals. Relative benefit and number-needed-to-treat (NNT), and relative risk and number-needed-to-harm (NNH) were calculated, with sensitivity analyses where appropriate to investigate differences between individual drugs and aspects of trial design.Twenty-six double blind placebo controlled trials had information from 2,853 patients for evaluation of efficacy. Topical NSAID was significantly better than placebo in 19 of the 26 trials, with a pooled relative benefit of 1.6 (95% confidence interval 1.4 to 1.7), and NNT of 3.8 (95% confidence interval 3.4 to 4.4) compared with placebo for the outcome of half pain relief at seven days. Results were not affected by outcome reported, or condition treated, but smaller trials yielded a larger estimate of efficacy. Indirect comparisons of individual topical NSAIDs showed that ketoprofen was significantly better than all other topical NSAIDs, while indomethacin was barely distinguished from placebo. Three trials, with 433 patients, compared topical with oral NSAID (two trials compared the same drug, one compared different drugs) and found no difference in efficacy. Local adverse events, systemic adverse events, or withdrawals due to an adverse event were rare, and no different between topical NSAID and placebo.Topical NSAIDs were effective and safe in treating acute painful conditions for one week.