OBJECTIVE: To evaluate the efficacy and safety of selective serotonin re-uptake inhibitors (SSRIs) in the treatment of premature ejaculation (PE). METHODS: From MEDLINE (Jan, 1950-Mar, 2008), EMBASE (Jan, 1980-Mar, 2008), The Cochrane Library (Issue 1, 2008) and CNKI (Jan, 1979-Mar, 2008), we retrieved and screened the randomized controlled trials (RCT) and randomized crossover trials (RT) as well as various related data, published and unpublished, on the treatment of PE with SSRIs. The methodological quality of the included trials was evaluated by 2 reviewers. Meta-analyses were conducted with RevMan 5.0 on the homogeneous studies. RESULTS: Totally 22 studies on 4 291 patients were included. Meta-analyses showed that after treated with sertraline, fluoxetine, paroxetine, citalopram, dapoxetine and fluvoxamine, the WMD (95% CI) values of the changes in intravaginal ejaculatory latency time (IELT) were 2.63 (1.80, 3.46), 2.21 (1.50, 2.92), 4.31 (2.71, 5.91), 3.82 (3.39, 4.25), 1.57 (1.31, 1.84) and 0.01 (0.71, 0.73) respectively; the RR (95% CI) values of the sexual satisfaction rate of the patients were 1.65 (1.12, 2.43), 2.93 (0.50, 17.31), 3.08 (2.27, 4.17), 2.48 (1.99, 3.09) and 2.93 (2.36, 3.65), and those of their partners were 1.47 (0.98, 2.21), 2.88 (0.38, 21.77), 4.81 (3.15, 7.36), 5.38 (3.75, 7.72) and 2.91 (1.09, 7.78) respectively for sertraline, fluoxetine, paroxetine, citalopram and dapoxetine. CONCLUSION: All the known SSRIs but fluvoxamine could prolong IELT, and some could improve the sexual satisfaction of both the patients and their partners, but their adverse effects should be noted. The moderate possibility of selection bias and publication bias in the included studies might have a negative impact on the evidence intensity of our results. We expect more reliable evidence from more randomized controlled trials.