This network meta-analysis aimed at assessing the influence of tramadol on the intravaginal ejaculatory latency time (IELT) and sexual satisfaction score (SSS) in treating patients with premature ejaculation (PE). The PubMed, Embase, Cochrane Library databases (until July 2021), and original references of the included articles was systematically retrieved. The PRISMA checklist was followed. Finally, 14 articles including 1971 patients were included in this analysis. The results indicated that patients who were treated with tramadol (50 mg, 62 mg, 89 mg, and 100 mg) were superior to those treated with placebo in terms of IELT (p = .003, p < .00001, p < .00001, and p < .00001, respectively), but 25 mg tramadol did not show a significant advantage (p = .06). Patients who were treated with tramadol (50 mg and 100 mg) had a better efficacy than who were treated with 25 mg tramadol in the IELT (p < .00001 and p < .00001), but the effect of 50 mg tramadol and 100 mg tramadol were not significantly different (p = .17). The tramadol group had the better effect than the placebo group in the SSS (p < .0001). And 50 mg tramadol showed a significant improvement compared with 20 mg paroxetine, as assessed by the IELT (p = .03) and SSS (p = .03). Safety assessments including adverse events suggested that tramadol was well tolerated. Tramadol showed a better improvement of IELT and SSS than placebo or paroxetine, and 50 mg tramadol may be a more reasonable therapeutic dose for patients with PE.