Objective To systematically evaluate the effectiveness and accuracy of brain natriuretic peptide (BNP) for predicting postoperative cardiovascular events of non-cardiac surgery. Methods Databases including The Cochrane Library, PubMed, Ovid, EMbase, WanFang Data and CNKI were searched electrically to collect literature published from 2000 to 2011, and relevant periodicals and references of the included studies were also manually retrieved. According to the inclusion and exclusion criteria, related cohort studies were selected, data were extracted, and quality of the included studies was evaluated by two reviewers independently. Then meta-analysis was conducted using RevMan 5.0 software. Results A total of 11 studies involving 3 649 patients were included. The results of meta-analysis showed that, compared with patients with lower BNP levels than the cut-off point before surgery, patients with higher BNP levels than the cut-off point before surgery suffered from a higher incidence of cardiovascular events, with a significant difference (OR=27.54, 95%CI 17.49 to 43.35, Plt;0.000 01), while the result of N-terminal pro-brain natriuretic peptide (NT-proBNP) was similar to that of BNP (OR=19.53, 95%CI 13.54 to 28.17, Plt;0.000 01). Conclusion Postoperative higher levels of BNP and NT-BNP can be used to predict postoperative cardiovascular events of non-cardiac surgery patients. This conclusion needs to be further proved by more high quality studies due to the quality limitation of the included studies.