BACKGROUND: Use of antipsychotic medications has been associated with increased risk of cerebrovascular events; however, this association remains questionable given conflicting evidence in the literature. OBJECTIVES: We conducted a systematic review and meta-analysis to determine the risk of stroke with the use of antipsychotic medications.  METHODS: All articles published between 1970 and February 2015 were identified by comprehensively searching PubMed, MEDLINE and EMBASE without language restrictions. Observational studies comparing stroke outcomes in antipsychotic patients with non-users were selected. Two authors independently extracted study characteristics and indicators of study quality. Newcastle-Ottawa Scale was adopted to assess risk of bias. Pooled odd ratios (ORs) and heterogeneity (I) were estimated on the basis of random effects models. RESULTS: We identified 22 potentially relevant studies from 1,171 citations. Of these, 9 studies (3 cohort, 5 case–control and 1 case-case-time-control) with a total of 155,789 subjects and 10,203 cases of stroke were eligible for final analysis. Use of antipsychotics was associated with a significantly higher risk of developing stroke [OR 1.57, 95% confidence interval (CI) 1.29-1.98, I= 92.4%]. The pooled OR for stroke was 1.58 [95% CI 1.01-2.49, I = 68.4%] with exposure to conventional antipsychotics and 1.06 [95% CI 0.59-1.89, I = 56.2%] with exposure to atypical antipsychotics. Subgroup analysis of conventional antipsychotics showed elderly patients over 64 years old were at lower risk for stroke [OR 1.37, 95% CI 0.87-2.17, I = 64.5%]. Due to limited data on individual agents, only Risperidone was evaluated in the subgroup analysis of atypical antipsychotics. Risperidone users were less likely to develop stroke than non-users of antipsychotics [OR 0.63, 95% CI 0.33-1.17, I = 55.2%]. CONCLUSIONS: Exposure to conventional antipsychotic was associated with a significant increase in stroke risk. Nonetheless, use of atypical antipsychotics revealed lower risk of stroke. Given heterogeneity among eligible studies, additional research is needed.