Abstract Chondrichthyans are highly vulnerable to fisheries overexploitation, and postcapture mortality poses a significant threat to most species. Global bycatch mitigation guidelines recommend adopting hierarchical decision‐making approaches tailored to species‐specific vulnerabilities and socioeconomic and regulatory contexts. Effective implementation of such strategies requires robust understanding of the factors driving vulnerability to postcapture mortality. To address this need, we developed a machine learning method to identify key drivers of at‐vessel mortality (AVM) based on a broad set of biological, environmental, and fishing‐related parameters. We sought to reveal interactions among predictors, nonlinear responses between these variables and mortality risk, and threshold values beyond which the likelihood of mortality increased markedly. We applied this approach to trawl bycatch data on small‐spotted catshark ( Scyliorhinus canicula ) and blackmouth catshark ( Galeus melastomus ) in the western Mediterranean. Body size, air temperature, and on‐deck time emerged as the primary AVM drivers. Mortality risk increased substantially at temperatures above 20°C for S. canicula and 16°C for G. melastomus , with on‐deck exposure exceeding 15 min, and when body size was below 40 and 55 cm, respectively. Identification of these drivers and thresholds provides valuable insights for bycatch mitigation; can inform strategies for more threatened, closely related, or physiologically and ecologically similar species; and may support management authorities in adopting targeted bycatch avoidance strategies, gear selectivity, and mortality reduction measures. Such measures can be tailored to specimens, areas, and periods of heightened mortality risk to maximize effectiveness. Furthermore, our scalable modeling approach offers a robust tool for identifying critical AVM drivers across regions and species, and its applicability can be extended to broader fisheries management and global conservation efforts.