Background: Early discontinuation of attention-deficit/hyperactivity disorder (ADHD) medication is common and linked to worse outcomes. Identifying clinical predictors could aid personalised treatment yet evidence is inconsistent across ages and countries/regions. Objective: Investigate psychiatric and neurological comorbidity as predictors of early ADHD medication discontinuation in new ADHD medication users across age groups, sex and countries/regions. Methods: Using health records from eight countries/regions, we identified 1 000 411 (44% female) new ADHD medication users (2011–2020). Discontinuation was defined as a ≥180 day gap between dispensations. We examined 23 indicators of psychiatric or neurological comorbidity, severity and psychotropic medication use. Associations were estimated using Cox regression, pooled with random-effects meta-analyses and stratified by age-at-initiation and sex. Findings: Discontinuation rates varied widely (children 19%–61%, adolescents 37%–68%, young adults 52–67%, adults 38%–68%). In pooled analyses, earlier discontinuation in children was predicted by intellectual disability, autism and use of psychotropic medications (HR range 1.32–1.51), while conduct/oppositional defiant disorder (CD/ODD) was protective (HR 0.83, 95% CI 0.73 to 0.94). In adolescents, no indicators remained statistically significant after multiple-testing control. In young adults, CD/ODD (HR 1.42, 95% CI 1.30 to 1.55), and in adults, schizophrenia (HR 1.25, 95% CI 1.09 to 1.44) and tic disorders (HR 1.27, 95% CI 1.11 to 1.46) predicted earlier discontinuation. Statistical heterogeneity was substantial, largely driven by US estimates. In meta-analyses excluding the USA, additional associations emerged. For example, in children, OCD and anxiety disorders predicted earlier discontinuation, while eating disorders and antidepressants/anxiolytics were protective in adults. Associations with schizophrenia, tic disorders and CD/ODD were no longer significant. Country-specific analyses showed similar association patterns, except in the USA, Hong Kong and the UK. Sex differences were limited. Conclusions: Children with neuropsychiatric comorbidity and related comedication are more likely to discontinue ADHD medication early, whereas few consistent predictors were seen from adolescence onwards. Marked cross-country variation, particularly in the USA, points to system-level influences on treatment patterns. Clinical implications: Improving ADHD medication persistence will require consideration of healthcare context and age-specific strategies, including close monitoring for children with complex neuropsychiatric profiles, and consideration of broader factors in adolescents and adults, where clinical predictors were limited.