Objective To update, synthesise and provide meta-analytical evidence of the associations between domain-specific physical activity (PA) and mental health and mental ill-health outcomes. Design Systematic review and multilevel meta-analysis. Data sources In March 2024, we systematically searched five databases. Eligibility criteria Methods employed replicated those of a previous review in 2017. All studies examining associations between domain-specific PA and specified mental health outcomes were included. Results 372 studies met inclusion criteria and 361 were included in the meta-analysis. Across the 372 studies (combined sample size of 3 323 711), 338 examined leisure-time PA, 54 work-related PA, 72 transport-related PA, 44 household PA, 5 school sport and 8 physical education. Multilevel meta-analyses showed that leisure-time PA (r=0.205, 95% CI 0.157 to 0.253), transport-related PA (r=0.138, 95% CI 0.042 to 0.231) and household PA (r=0.096, 95% CI 0.025 to 0.165) were positively associated with mental health . Leisure-time PA (r=−0.149, 95% CI −0.189 to –0.11) and school sport (r=−0.096, 95% CI −0.115 to –0.077) were inversely associated with mental ill health . However, work-related PA (r=0.134 95% CI 0.069 to 0.199) was positively associated with mental ill health . Conclusion The direction of the association between PA and mental health/mental ill health is dependent on the domain in which PA occurs. Promoting PA for leisure purposes is likely to yield the greatest benefits for both promoting mental health and preventing mental ill health. As such, leisure-time PA should be prioritised in messaging, guidelines and interventions/programmes designed to support mental health through PA. PROSPERO registration number CRD42024510303. Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. Data are viewable in [online supplemental file 5][1]. All other data are available upon reasonable request. [1]: #DC5