Background Several small sample-size observational studies evaluated the association of plasma brain natriuretic peptide (BNP) or N-terminal pro-brain natriuretic peptide (NT-proBNP) with atrial fibrillation (AF),but the results were contradictory.We aimed to perform a meta-analysis of relevant studies to evaluate the availability of this association.Methods We performed an extensive literature search on PubMed,Web of Science (WOS) and the Cochrane Library databases.Pooled standardized mean difference (SMD) and 95％ confidence interval (CI) were calculated to assess the strength of association using random effects models.We performed sensitivity and subgroup analyses to explore the potential sources of heterogeneity.We also estimated publication biases.Statistical analyses were performed using the STATA 12.0 software.Results A total of 11 studies including 777 cases and 870 controls were finally analyzed.Overall,the brain natriuretic peptide/N-terminal pro-brain natriuretic peptide levels were higher in atrial fibrillation patients than controls without atrial fibrillation.Results showed that the SMD in the natriuretic peptide levels between cases and controls was 2.68 units (95％CI 1.76 to 3.60); test for overall effect z-score=5.7 (P ＜0.001).There was significant heterogeneity between individual studies (I2=97.8％; P ＜0.001).Further analysis revealed that differences in the assay of natriuretic peptide possibly account for this heterogeneity.Conclusions Increased BNP/NT-proBNP levels were associated with the presence of atrial fibrillation.This finding indicates that BNP/NT-proBNP may prove to be a biomarker of an underlying predisposition to AF.