To evaluate the efficacy of manifold ovarian stimulation protocols for patients with poor ovarian response. PubMed, Embase, Cochrane Library and Web of Science were systematically searched until February 14, 2021. Primary outcomes included clinical pregnancy rate per initiating cycle and low risk of cycle cancellation. Secondary outcomes included number of oocytes retrieved, number of metaphase II (MII) oocytes, number of embryos obtained, number of transferred embryos, endometrial thickness on triggering day and estradiol (E2) level on triggering day. The network plot, league table, rank probabilities and forest plot of each outcome measure were drawn. Therapeutic effects were displayed as risk ratios (RRs) or mean differences (MDs) with 95% confidence intervals (CIs). This network meta-analysis included 15 trials on 2173 participants with poor ovarian response. Delayed start GnRH antagonist was the best regimen in terms of clinical pregnancy rate per initiating cycle (74.04% probability of being the optimal), low risk of cycle cancellation (75.30%), number of oocytes retrieved (68.67%), number of metaphase II (MII) oocytes (97.98%) and endometrial thickness on triggering day (81.97%), while for E2 level on triggering day, microdose GnRH agonist (99.25%) was the most preferred. Regarding number of embryos obtained and number of transferred embryos, no statistical significances were found between different ovarian stimulation protocols. Delayed start GnRH antagonist and microdose GnRH agonist were the two superior regimens in the treatment of poor ovarian response, providing favorable clinical outcomes. Future investigation is needed to confirm and enrich our findings.