Background Immune checkpoint inhibitors have transformed melanoma therapy but frequently cause immune-related adverse events (irAEs), including colitis, that limit treatment. Reliable biomarkers predicting toxicity remain lacking. Methods In this retrospective, multicenter study, we analyzed pretreatment serum samples from 331 patients with metastatic melanoma treated with anti-CTLA-4 (ipilimumab), anti-PD-1 (pembrolizumab or nivolumab), or combination ipilimumab/nivolumab. IgG autoantibody reactivity against 832 human protein antigens, including autoimmune targets, cytokines, tumor-associated antigens, and cancer pathway proteins, was profiled using multiplex bead-based arrays. Statistical analysis (Significance Analysis of Microarrays and Cox regression) identified autoantibody signatures associated with subsequent irAEs and immune-related colitis (ir-colitis). Results We detected 47 autoantibodies predictive of irAEs, with KRT7, RPLP2, UBE2Z, and GPHN emerging as the strongest markers. Anti-KRT7 and anti-GPHN were specifically predictive in patients receiving PD-1 monotherapy, whereas anti-RPLP2 was associated with irAEs in ipilimumab/nivolumab combination therapy. For ir-colitis, 38 autoantibodies were identified, with five (PIAS3, RPLP0, UBE2Z, KRT7, and SDCBP) showing consistent predictive value across treatment groups. Anti-PIAS3 and anti-RPLP0 increased ir-colitis risk, while anti-SDCBP conferred protection. Notably, predictive profiles differed between PD-1-based and CTLA-4-based regimens, underscoring divergent mechanisms of toxicity. Several autoantibodies predictive of irAEs or ir-colitis also correlated with clinical outcome. ATG4D, MAGEB4, and IL4R were associated with prolonged progression-free and overall survival, whereas FGFR1 predicted both reduced irAE risk and inferior survival, consistent with the link between heightened immune activation, toxicity, and therapeutic benefit. Conclusions This study, to our knowledge, is the largest pretreatment autoantibody screen in melanoma immunotherapy, demonstrates that serum autoantibody profiles can stratify patients at risk for irAEs and ir-colitis. The identified signatures connect tumor-related and immunity-related antigens, stress-response pathways, and autoimmune mechanisms. Pretreatment autoantibody profiling offers a promising biomarker-driven approach for individualizing risk assessment, improving patient selection, and guiding early intervention strategies to enhance the safety of immune checkpoint blockade in melanoma. Beyond toxicity prediction, our findings also suggest that specific autoantibodies may reflect underlying immune activation states linked to therapeutic response.