PURPOSE: The objective of this study was to identify potential predictors of immune-related adverse events (irAEs) in cancer patients receiving immune checkpoint inhibitor therapy among serum indexes, case data, and liquid biopsy results. METHODS: We retrospectively analyzed 418 patients treated with anti-programmed cell death 1(PD-1)/PD-1 ligand (PD-L1) inhibitors from January 2018 to May 2022 in our cancer center. We identified factors that correlated with the occurrence of irAEs and evaluated associations between irAEs and anti-PD-1/PD-L1 inhibitor responses. RESULTS: The incidence of irAEs was 42.1%, and pneumonitis (9.1%), thyroid toxicity (9.1%), cardiotoxicity (8.1%), and dermatologic toxicity (6.9%) were the four most common irAEs. Multivariate logistic analysis identified female sex, antibiotic use, higher post-treatment neutrophil-to-lymphocyte ratio (NLR), and higher baseline circulating tumor cell (CTC) level, as predictive biomarkers for the occurrence of irAEs. A lower baseline prognostic nutritional index (PNI), body mass index (BMI) ≥ 25 kg/m, and higher post-treatment lactate dehydrogenase (LDH) level were predictive factors for more severe irAEs (higher severity grade). Patients without irAEs had better overall survival than those with irAEs. Specifically, pneumonitis and cardiotoxicity were found to be significant predictors of poor prognosis in the irAE subgroup with different organ-related irAEs. Low-dose steroid (dexamethasone 10 mg) treatment had no significant effect on outcomes. CONCLUSIONS: Gender, antibiotic use, post-treatment NLR, and baseline CTC level are potential predictive biomarkers of irAEs, while baseline PNI, BMI, and post-treatment LDH may predict the severity of irAEs. The predictive effect of irAE occurrence on survival benefit may depend on the type of irAE.