Anti-PD-1/PD-L1 inhibitors provide a survival advantage over conventional therapies for treatment of advanced or metastatic cancer. However, the factors determining which patients benefit the most from anti-PD-1/PD-L1 inhibitors are unknown, making treatment-related decisions difficult. We performed a systematic review and meta-analysis of acquired data to assess the efficacy and toxicity of anti-PD-1/PD-L1 inhibitors in advanced and metastatic cancer. A thorough search strategy was applied to identify randomised controlled trials (RCTs) in Pubmed, Embase, Cochrane, and major conferences. Studies meeting predefined selection criteria were selected, and two independent investigators performed data extraction; overall survival (OS), progression-free survival (PFS), and overall response rate were compared between anti-PD-1/PD-L1 inhibitors and control therapies. We calculated the pooled response rate and 95% CIs of all-grade and high-grade (≥3) adverse effects and evaluated the within-study heterogeneity using subgroup, sensitivity, and meta-regression analyses. In final, we included eligible 35 RCTs (21047 patients). The main estimated hazard ratios (HRs) for OS and PFS were 0.76 (0.71-0.82) and 0.81 (0.73-0.89) in a random-effects model. The anti-PD-1/PD-L1 inhibitor group had a significantly high risk for all-grade immune-related adverse events. Anti-PD-1/PD-L1 inhibitors were identified as a preferable treatment option for advanced or metastatic cancer patients who are male, aged < 65 years, current or former smokers, had no CNS or liver metastasis, had not EGFR mutation, and had high PD-L1 expression.