Positional, functional and association studies have strongly implicated the dystrobrevin binding protein 1 gene (DTNBP1) as a promising novel candidate gene for schizophrenia. Since the first association study was reported, there have been many attempts to replicate it. However the results have been mixed and these subsequent studies have produced negative as well as positive results. To reconcile these conflicting findings and to give a comprehensive picture of the relationship of DTNBP1 and schizophrenia, the current meta-analysis combined all published association studies involving nine polymorphisms up to May 2006. The results (12 studies including 3429 cases, 3376 controls and 721 trios) showed that there were five single nucleotide polymorphisms (SNPs) with p values < 0.05, however, sensitivity analyses showed that only one SNP was consistent across all nine studies (four of the five SNPs became non-significant after removal of one study), indicating that one study may cause the association findings for each of these four SNPs. In conclusion, there is only a weak association of one SNP in DTNBP1 with schizophrenia, which is not significant after multiple testing.