A recent report showed significant associations between several SNPs in a previously unknown EST cluster with schizophrenia. 1 The cluster was identified as the human dystrobrevin binding protein 1 gene (DTNBP1) by sequence database comparisons and homology with mouse DTNBP1. 2 However, the linkage disequilibrium (LD) among the SNPs in DTNBP1 as well as the pattern of significant SNP–schizophrenia association was complex. This raised several questions such as the number of susceptibility alleles that may be involved and the size of the region where the actual disease mutation (s) could be located. To address these questions, we performed different single-marker tests on the 12 previously studied and 2 new SNPs in DTNBP1 that were re-scored using an improved procedure, and performed a variety of haplotype analyses. The sample consisted of 268 Irish multiplex families selected for high density of schizophrenia. Results suggested a simple structure where the LD in the target region could be explained by 6 haplotypes that together accounted for 96% of haplotype diversity in the whole sample. From these six, a single high-risk haplotype was identified that showed a significant association with schizophrenia and explained the pattern of significant findings in the analyses with individual markers. This haplotype was 30 kb long, had a large effect, could be measured with two tag SNPs only, had a frequency of 6% in our sample, seemed to be of relatively recent origin in evolutionary terms, and was equally distributed over Ireland. Implications of these findings for follow-up and replication studies are discussed.