The effect of erythropoietin (EPO) on the prevention of cardiac surgery-associated acute kidney injury (CSA-AKI) is controversial. Therefore, we undertook the meta-analysis of randomized controlled trials (RCTs) to assess the efficacy and safety of EPO on the prevention of CSA-AKI in adult patients and to explore whether risk factors for AKI could explain the inconsistent effects. PubMed and EMbase databases were searched to identify eligible RCTs. The meta-analysis was performed with fixed- or random-effects models according to the heterogeneity, and the subgroup analysis stratified by risk factors for AKI was carried out. Five RCTs involving 423 patients were included. Overall, EPO administration was not associated with a reduced incidence of CSA-AKI [relative risk (RR): 0.64, 95% confidence interval (CI): 0.35-1.16], with a moderate heterogeneity (I(2) = 67.4%, heterogeneity P = 0.02). Subgroup analysis showed that, in patients without high risk factors for AKI, EPO administration could significantly reduce the incidence of CSA-AKI (RR: 0.38, 95% CI: 0.24-0.61), intensive care unit length of stay [standardized mean difference (SMD): -0.54, 95% CI: -1.05 to -0.04] and hospital length of stay (SMD: -0.48, 95% CI: -0.94 to -0.02). The test of heterogeneity was not significant in the two subgroups. EPO administration could significantly reduce the incidence of CSA-AKI, but not in patients with high risk factors for AKI. Substantial heterogeneity across trials could be attributed to high risk factors for AKI. However, our findings should be interpreted cautiously because of the limited studies included, and high-quality RCTs are warranted.