Background Impairment of prospective memory (PM) in schizophrenia has gained increasing attention. This meta-analysis systematically examined PM impairment in schizophrenia. Methods Both English (PubMed, PsycINFO, EMBASE, and Cochrane Library) and Chinese (WanFang, Chinese Biomedical and China Journal Net databases) databases were systematically searched from their inception until August 14, 2017. Case-control studies of PM in schizophrenia were included. Standardized mean differences (SMDs) and their 95% confidence interval (CI) were calculated using the random-effects model. Results Twenty-nine case-control studies (n = 2492) were included in the analyses. The overall and three subtypes of PM were compared between patients with schizophrenia (n = 1284) and healthy controls (n = 1208). Compared to healthy controls, patients performed significantly poorer in overall (SMD = −1.125), time-based (SMD = −1.155), event-based (SMD = −1.068), and activity-based PM (SMD = -0.563). Subgroup analyses revealed significant differences between older and younger patients (SMD = -1.398 vs. -0.763), higher male predominance and no sex predominance (SMD = -1.679 vs. -0.800), lower and higher education level (SMD = -1.373 vs.-0.637), chronic and first-episode patients (SMD = -1.237 vs. -0.641) and between eco-valid and dual-task laboratory measurements (SMD = -1.542 vs. -0.725) regarding overall PM. Meta-regression analysis showed that higher negative symptom score was significantly associated with more severe overall PM impairment in patients (P = 0.022). Conclusions In this meta-analysis the overall PM and all its subtypes, particularly the time-based PM, were significantly impaired in schizophrenia.