Aims: Binge drinking at age 22 can be predicted from adolescent brain structure at age 14 with an accuracy above 70 %. We investigated whether psychometric characteristics such as binge- and addiction-related measures, psychiatric comorbidities, or personality characteristics explain this accurate prospective prediction.Design: As part of the prospective longitudinal IMAGEN project, participants underwent structural brain imaging at age 14, binge drinking assessments at ages 14 and 22, and psychometric questionnaire assessments at ages 14 and 22.Setting: France, Germany, Ireland, United Kingdom.Participants: Individuals with no lifetime binge drinking episodes at age 14 (n = 555, 47.38 % female).Measurements: Brain structural features were derived from T1-weighted magnetic resonance and diffusion tensor imaging. Binge drinking was assessed using the ESPAD questionnaire. We employed machine learning to predict binge drinking at age 22 from brain structure at age 14 while controlling for psychometric variables using counterbalancing with oversampling. Each variable’s explanatory role was quantified by the reduction in accuracy after controlling for each variable.Findings: The inclusion of eating-related variables, such as uncontrolled eating (TFEQ-R18), led to the largest reductions in the accuracy of the prospective prediction of binge drinking from brain structure (max. -6.6%). To a less consistent extent, other binge- and addiction-related measures (max. -5.6%), psychiatric comorbidities (max. -6.3%), and personality characteristics (max. -4.1%) resulted in a lower prediction accuracy when controlled for.Conclusions: Eating behavior and other binge- and addiction-related characteristics partially explain the predictability of binge drinking from brain structure in individuals with no previous binge drinking experiences, while other factors that previously had been related to binge drinking, such as stress and impulsivity, do not. Binge- and addiction-related behaviors may share similar underlying risk factors, which might facilitate early detection of high risk for binge drinking and inform trans-diagnostic prevention approaches.