92 Background: Immunotherapy (IO) can lead to immune-related adverse events (irAEs). Evidence on the association of irAEs and efficacy is limited. Methods: We conducted a systematic review in MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials (inception to July 1, 2019) to identify randomized controlled trials (RCTs) reporting irAEs, incidence and efficacy data for pembrolizumab (PEM), nivolumab (NIVO), ipilimumab (IPI), atezolizumab, avelumab, durvalumab, and aldesleukin in lung, renal, head and neck cancer, and melanomas. RCTs assessing IO monotherapy or IO combinations in at least one arm were included. Evaluated outcomes were 1) irAE incidence (rash, pruritis, diarrhea, colitis, hypothyroidism, hyperthyroidism, transaminitis, hypophysitis, pneumonitis, arthralgia, anemia, and hepatitis); and 2) efficacy (response and survival). irAE incidence data will be pooled and meta-analysis performed to assess the association of irAEs with efficacy; heterogeneity between RCTs will be evaluated. Results: Fifty RCTs were included: IOs were compared to chemotherapy or chemotherapy-combinations (19), to other interventions (8), to placebo (3), and head-to-head (20). irAE reporting and definitions were heterogeneous across RCTs. The most common all-grade irAEs were skin, GI, and endocrine. For skin irAEs, rash ranged from 0% (PEM 2 mg/kg, n = 6) to 73% (NIVO + IPI, n = 11); pruritus was from 1% (PEM 2 mg/kg, n = 89) to 50% (NIVO + IPI, n = 6). For GI irAEs, diarrhea ranged from 0% (PEM 2 mg/kg, n = 6) to 64% (NIVO + IPI, n = 11); colitis was from 0% (NIVO 3 mg/kg, n = 98) to 23% (NIVO + IPI, n = 94). For endocrine irAEs, hypothyroidism ranged from 0% (NIVO 3 mg/kg, n = 12) to 83% (NIVO + IPI, n = 6), hyperthyroidism was from 0% (PEM 2 mg/kg, n = 6 and IPI 3 mg/kg, n = 46) to 27% (NIVO + IPI, n = 11), and hypophysitis was from 0% (PEM 10 mg/kg, n = 84 and NIVO 3 mg/kg, n = 25) to 26% (NIVO + IPI, n = 35). Liver, pulmonary, and musculoskeletal irAEs, and anemia, were reported less frequently and with lower incidence. Conclusions: irAEs are increasingly reported in IO RCTs, but lack reporting and definition consistency. The meta-analysis results may provide clarity on irAEs incidence and association with efficacy.