Background: Many immuno-oncology (IO) trials are conducted without biomarker selection. We performed a meta-analysis of phase I/II clinical trials evaluating immune checkpoint inhibitors (ICIs) to determine the association between biomarkers and clinical outcomes, if any. Methods: We searched PubMed for phase I/II clinical trials of drugs approved by the Food and Drug Administration (labeled, off-label, or combined with investigational ICIs or other treatment modalities) from 2018 to 2020. We compared the objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) between biomarker-positive and biomarker-negative groups, using data from correlative studies. Results: Overall, 174 clinical studies that included 19,178 patients were identified and 131 studies investigated >30 correlative biomarkers, that included PD-L1 expression (≥1%, 111 studies), tumor mutational burden (20 studies), and microsatellite instability/mismatch repair deficiency (10 studies). Overall, 123, 46, and 30 cohorts (drugs, tumor types, or biomarkers) with 11,692, 3,065, and 2,256 patient outcomes for ORR, PFS, and OS, respectively, were analyzed in correlation with biomarkers. Meta-analyses demonstrated that ICIs in patients with biomarker-positive tumors were associated with higher ORR (odds ratio 2.15 [95% CI, 1.79-2.58], p<0.0001); and longer PFS (hazard ratio [HR] 0.55 [95% CI, 0.45-0.67], p<0.0001), and OS (HR 0.65 [95% CI, 0.53-0.80], p<0.0001) compared with those with biomarker-negative tumors. Significance for ORR and PFS was retained in multivariate analysis (p<0.001) (OS, not included owing to the small number of trials reporting OS). Conclusions: Our data suggest that IO biomarkers should be used in patient selection for ICIs. Prospective studies are warranted.