Objective To assess the efficacy of topical non-steroidal anti-inflammatory drugs （NSAIDs） in the treatment of osteoarthritis （OA）. Methods MEDLINE, EMBASE, Scientific Citation Index, CINAHL, The Cochrane Library, CBMdisc and abstracts from conference were searched from 1966 to March 30, 2005. Randomized controlled trials （R.CT） comparing topical non-steroidal anti-inflammatory drug （NSAIDs） with placebo or oral NSAIDs in OA were induded. Effect size （ES） was calculated for pain, function and stiffness. Relative risk （RR） was calculated for dichotomous data such as clinical response rate and adverse effect rate. Number needed to treat to obtain the clinical response was estimated. The quality of trials was assessed and sensitivity analyses were undertaken. Results Topical NSAIDs were superior to placebo in relieving pain due to osteoarthritis only in the first 2 weeks of treatment; ES （95% CI） were 0.41 （0. 16 to 0.66） and 0.40 （0.15 to 0.65） at week 1 and 2 respectively. However, the effects were short-lived and no benefit was observed over placebo at the third and fourth week. A similar pattern was observed with function, stiflhess and clinical response RR and number needed to treat. Topical NSAIDs were inferior to oral NSAIDs at week 1, and associated with more local side effects such as rash, itch or burning （RR 5.29, 95% CI 1.14 to 24. 51 ）. Conclusions Only very shortterm （less than 4 weeks） RCTs have assessed topical NSAID efficacy in OA ; after 2 weeks no efficacy above placebo has been obsevrved. There are no trial data to support the long-term use of topical NSAIDs in osteoarthritis.