Evidence that a gene or genes on chromosome 22 is involved in susceptibility to schizophrenia comes from two sources: the increased incidence of schizophrenia in individuals with 22q11 deletion syndrome (22q11DS) and genetic linkage studies. In mice, hemizygous deletion of either Tbx1 or Gnb1l can cause deficits in pre-pulse inhibition, a sensory motor gating defect which is associated with schizophrenia. We tested the hypothesis that variation at this locus confers risk of schizophrenia and related disorders in a series of case-control association studies. First, we found evidence for a male-specific genotypic association (P = 0.00017) TBX1/GNB1L in 662 schizophrenia cases and 1416 controls from the UK. Moreover, we replicated this finding in two independent case-control samples (additional 746 cases and 1330 controls) (meta analysis P = 1.8 x 10(-5)) and also observed significant evidence for genotypic association in an independent sample of 480 schizophrenia parent-proband trios from Bulgaria with markers at this locus, which was again strongest in the male probands (P = 0.004). Genotyping the most significant SNPs in a sample of 83 subjects with 22q11DS with and without psychosis again revealed a significant allelic association with psychosis in males with 22q11DS (P = 0.01). Finally, using allele specific expression analysis, we have shown that the markers associated with psychosis are also correlated with alterations in GNB1L expression, raising the hypothesis that the risk to develop psychosis at this locus could be mediated in a dose sensitive manner via gene expression. However, other explanations are possible, and further analyses will be required to clarify the correct functional mechanism.