Objective To evaluate efficacy and safety of topical tacrolimus（FK506）for atopic dermatitis. Methods Randomized controlled trials (RCTs) were identified from specialized trials registered in Cochrane Skin Group (July, 2003), the Cochrane Library (issue 2, 2003), Medline (1996-2003), Embase (1984-2003) and CBM (1978-2003). We handsearched the published and unpublished data and Cochrane Skin Group 8th Annual Meeting. RCTs comparing tacrolimus with placebo or hormone were included. Data were extracted and evaluated by two reviewers independently. Results Eight randomized controlled trials involving 4 122 patients were included, with all trials of high methodological quality. Meta-analysis indicated that 0.03% tacrolimus was more effective than placebo, 1% hydrocortisone acetate and 0.1% hydrocortisone butyrate with odds ratio of 3.03 [95%CI (1.05, 8.73), P=0.04], 0.1% tacrolimus was more effective than placebo, 1% hydrocortisone acetate and 0.1% hydrocortisone butyrate with odds ratio of 3.84 [95%CI (1.43, 10.32), P=0.008], 0.3% tacrolimus was more effective than placebo with odds ratio of 3.20 [95%CI (1.31, 7.79), P=0.01], the odds ratio of 0.1% tacrolimus vs 0.03% tacrolimus was 1.40 [95%CI (1.13, 1.72), P=0.002]. No serious adverse effects were identified. Conclusions Topical tacrolimus for atopic dermatitis is more effective than placebo and 1% hydrocortisone acetate. 0.1% tacrolimus is more effective than 0.03% tacrolimus. No conclusion could be drawn when tacrolimus is compared with 0.1% hydrocortisone butyrate. Tacrolimus tends to improve EASI scores, head and neck scores as well as HRQL scores, but more randomized controlled trials are necessary to draw definite conclusions.