The association between DTNBP1 genotype and cognitive abilities was investigated in three population samples (1054 Scottish, 1806 Australian and 745 English) of varying age. There was evidence in each of the cohorts for association (P < 0.05) to single nucleotide polymorphisms (SNPs) and haplotypes previously shown to relate to cognition. By comparison with previous findings, these associations included measures of memory, and there was at best equivocal evidence of association with general cognitive ability. Of the SNPs typed in all three cohorts, rs2619528 and rs1011313 showed significant association with measures of executive function in two cohorts, rs1018381 showed significant association with verbal ability in one cohort and rs2619522 showed significance/marginal significance with tests of memory, speed and executive function in two cohorts. For all these SNPs, the direction and magnitude of the allelic effects were consistent between cohorts and with previous findings. In the English cohort, a previously untested SNP (rs742105) located in a distinct haplotype block upstream of the other SNPs showed the strongest significance (P < 0.01) for measures of memory but weaker significance for general cognitive ability. Our results therefore support involvement of the dysbindin gene in cognitive function, but further work is needed to clarify the specific functional variants involved and the cognitive abilities with which they are associated.