BackgroundThere is uncertainty about the relationship between sleep duration and stroke risk. AimWe aimed to clarify the relationship between sleep duration and risk of stroke by using epidemiological evidence. MethodsWe searched MEDLINE and EMBASE to identify all studies that might be looking at the association between sleep duration and stroke, including both cohort and cross-sectional studies. Pooled hazard ratios (HRs) and odds ratios (ORs) were calculated through a random-effects model. ResultsOur study included a total of 12 cohort studies and 6 cross-sectional studies. Pooled results from the cohort studies showed that short sleep duration was associated with a higher risk for stroke [HR, 113; 95% confidence interval (CI) 102-125], and that long sleep duration also increases the risk of having a stroke (HR, 140; 95% CI, 116-164). Results from cross-sectional studies confirmed the relationship between stroke and inappropriate sleep duration, either too little sleep or too much. For short sleep duration, the OR was 171 (139-202); for long sleep duration, the OR was 212 (151-273). ConclusionBoth short and long sleep durations have a significant association with higher risk of stroke.