Purpose Considering the insufficient evidence supporting an ideal protocol for poor responder management in IVF/ICSI cycles, the aim of the current meta-analysis was to compare GnRH-antagonist versus GnRH-agonist protocols in poor responders, evaluating effectiveness and safety. Methods Meta-analysis was conducted using Medcalc 16.8 version software. Standardized mean differences (SMD), odds ratios (OR), and the respective 95% confidence intervals (CI) were determined appropriately. The Cochran Q statistic and the I 2 test were used to assess studies’ heterogeneity. Results GnRH-agonists were shown to correlate with fewer cancelled IVF/ICSI cycles ( p  = 0.044, OR = 1.268 > 1, 95% CI 1.007, 1.598), a larger number of embryos transferred ( p  = 0.008, SMD = − 0.230, 95% CI − 0.400, − 0.0599), and more clinical pregnancies ( p  = 0.018, OR = 0.748 < 1, 95% CI 0.588, 0.952). However, GnRH-antagonists resulted in a significantly shorter duration of ovarian stimulation ( p  = 0.007, SMD = − 0.426. 95% CI − 0.736, − 0.115). The number of oocytes and mature oocytes retrieved in both protocols did not differ statistically ( p  = 0.216, SMD = − 0.130, 95% CI − 0.337, 0.0763 and p  = 0.807, SMD = − 0.0203, 95% CI − 0.183, 0.142, respectively). Moreover, a high heterogeneity among studies was observed regarding duration of ovarian stimulation ( I 2  = 90.6%), number of oocytes ( I 2  = 82.83%)/mature oocytes retrieved ( I 2  = 70.39%), and embryos transferred ( I 2  = 72.83%). Conclusions Based on the present meta-analysis, agonist protocols could be suggested as a first choice approach, in terms of effectiveness. Due to the high studies’ heterogeneity, results should be considered with caution. Accordingly, larger cohort studies and meta-analyses like the present one will enhance the robustness of the emerging results to identify the ideal protocol for poor responders.