SummaryBackgroundComorbid mental illnesses and substance use disorders are associated with adverse criminal, social, and health outcomes. Yet, their burden is not reliably known among prison populations. We therefore aimed to estimate the prevalence of comorbid serious mental illnesses and substance use disorders (dual disorders) among people in prison worldwide.MethodsIn this systematic review and meta-analysis, we searched 15 electronic databases (ASSIA, CAB s, Criminal Justice Database, Embase, Global Health, Global Index Medicus, IBSS, MEDLINE, NCJRS, PAIS Index, PsycINFO, Russian Science Citation Index, Scielo, Social Services s, and Web of Science) and the grey literature (Open Grey and ProQuest Dissertations & Theses Global) for studies reporting the prevalence of serious mental illnesses and substance use disorders in prison populations published between Jan 1, 1980, and Sept 25, 2021, and contacted the authors of relevant studies. Empirical studies among unselected adult prison populations that applied representative sampling strategies and validated diagnostic instruments, and either reported the prevalence of dual disorders or had authors who could provide prevalence data in correspondence, were included. Two reviewers (GB and SDL) independently extracted data from the eligible studies; both current (up to 1 year) and lifetime prevalence were extracted, if available. We sought summary estimates. Our primary outcomes were comorbid non-affective psychosis with substance use disorders and comorbid major depression with substance use disorders. We conducted a random-effects meta-analysis, explored between-sample heterogeneity with meta-regression, and calculated odds ratios (ORs) to assess bidirectional relationships between mental and substance use disorders. Risk of bias was assessed by use of a standard tool. The study protocol was registered with PROSPERO, CRD42020207301.FindingsOf 11 346 records screened, we identified 34 studies reporting the prevalence of dual disorders among individuals in prison and received unpublished prevalence data for 16 studies, totalling 50 eligible studies and 24 915 people. The mean quality score of included studies was 7·8 (SD 1·2). We found that 3·5% (95% CI 2·2–5·0) had current non-affective psychosis with any comorbid substance use disorder, representing 443 (49·2%) of 900 people with non-affective psychosis, and 9·1% (5·6–13·3) had current major depression and comorbid substance use disorders, representing 1105 (51·6%) of 2143 people with major depression. Between-sample heterogeneity was high (I2>80%). People in prison with current non-affective psychosis were significantly more likely to have substance use disorders compared with those without (OR 1·7, 95% CI 1·4–2·2). People with major depression had higher odds of substance use disorders than those without (1·6, 1·3–2·0).InterpretationAround half of the prison population with non-affective psychosis or major depression have a comorbid substance use disorder. Consideration should be given to screening for dual disorders and implementing integrated and scalable treatments.FundingEconomic and Social Research Council, Agencia Nacional de Investigación y Desarrollo (Chile), and the Wellcome Trust.