BACKGROUND AND PURPOSE Urinary incontinence (UI) could be an indicator of increased mortality after new-onset stroke. The aim of the present meta-analysis was to characterize this association. METHODS A systematic search retrieved all studies exploring the post-stroke period and comparing death among patients suffering from UI with those without UI. Hazard ratios (HRs) were extracted or estimated from the published proportion of deaths. Various meta-analyses pooled unadjusted HRs, HRs adjusted for confounders and HRs stratified by subgroups of strokes (ischaemic or haemorrhagic), using models with random effects. Heterogeneity was explored through stratification of studies and meta-regression of predefined parameters. RESULTS The meta-analysis included 24 studies. UI increased the mortality among the general stroke patients in pooled unadjusted (HR, 5.1; 95% CI, 3.9-6.7) and adjusted (HR, 2.2; 95% CI, 1.8-2.7) analyses. This association was also found among ischaemic (HR, 8.5; 95% CI, 4.6-15.7) and haemorrhagic (HR, 3.9; 95% CI, 1.4-11.3) subgroups of strokes. Studies including indwelling catheters, published more than 10 years ago or with the highest quality on the selection criteria of the Newcastle-Ottawa Quality Assessment scale were associated with a greater effect of UI on mortality. Funnel plots showed a clear asymmetry for adjusted associations. After correcting for this potential publication bias, the pooled HRs still demonstrated a positive association between UI and mortality. CONCLUSIONS Urinary incontinence indicates high risk of death after a new-onset stroke. Validity of the analyses on adjusted models is limited by an obvious publication bias.