BACKGROUND: Second-generation antipsychotics (SGAs) are often prescribed in the treatment of Behavioral and Psychological Symptoms of Dementia (BPSD), however, their use has been discouraged in light of clinical trials suggesting that they cause an increased risk of cerebrovascular accidents (CVAs). OBJECTIVE: Aim of the study was to assess relative risk of CVA in dementia patients prescribed SGA rather than first-generation antipsychotics (FGAs), through meta-analysis of population-based studies. METHODS: A literature search was conducted using several relevant databases. Five studies were included in the review and data were pooled to conduct meta-analysis using the inverse variance method. RESULTS: A total of 79,910 patients were treated with SGAs and 1287 cases of CVA were reported. Of 48,135 patients treated with FGAs, a total of 511 cases of CVA were reported. The relative risk of CVA was 1.02 (95% CI 0.56-1.84) for the SGA group. There was no significant difference in the risk of stroke (p = 0.96) between groups, but significant heterogeneity was found among the results of included studies (p < 0.001). CONCLUSION: Meta-analysis of population-based data suggested that the use of SGAs as opposed to FGAs to control BPSD is not associated with significantly increased risk of CVA. Copyright © 2016 John Wiley & Sons, Ltd.