There are conflicting findings about prospective memory (PM) performance in bipolar disorder. This meta analysis systematically examined PM in patients with bipolar disorders. Articles were systematically searched in both English and Chinese databases, from their inception to Nov 15, 2016. Only case-control studies on PM in bipolar patients were included for analyses. The random effect model was used in all meta-analytic outcomes. Four studies (n = 390) comparing PM performance between patients with bipolar disorder (n = 208) and healthy controls (n = 182) were included. Three studies were rated as "high quality", while the quality of evidence in 3 meta-analyzable outcomes ranged from "moderate" (67%) to "high" (33%). Compared to healthy controls, bipolar disorder patients showed impairments in overall PM [2 studies, n = 196; SMD: - 1.08 (95%CI: - 1.61, - 0.55), P < 0.0001; I-2 = 65%], event-based PM [4 studies, n = 367; SMD: - 0.51 (95%CI: - 0.78, - 0.23), P = 0.0003; I-2 = 37%] and time-based PM performance [4 studies, n = 367; SMD: - 0.82 (95%CI: - 1.11, - 0.52), P < 0.0001; I-2 = 41%]. In this meta-analysis, both time-based PM and event-based PM deficits appeared to be evident in bipolar disorder.