UHMK1 has previously been implicated as a susceptibility gene for schizophrenia in the 1q23.3 region by significant evidence of allelic and haplotypic association between schizophrenia and several genetic markers at UHMK1 in a London-based case-control sample. Further fine mapping of the UHMK1 gene locus in the University College London schizophrenia case-control sample was carried out with tagging SNPs. Two additional SNPs were found to be associated with schizophrenia (rs6604863 P = 0.02, rs10753578 P = 0.017). Tests of allelic and haplotypic association were then carried out in a second independent sample from Aberdeen consisting of 858 individuals with schizophrenia and 591 controls. Two of these SNPs also showed association in the Aberdeen sample (rs7513662 P = 0.0087, rs10753578 P = 0.022) and several haplotypes were associated (global permutation P = 0.0004). When the UCL and Aberdeen samples were combined three SNPs (rs7513662 P = 0.0007, rs6427680 P = 0.0252, rs6694863 P = 0.015) and several haplotypes showed association (eg HAP-A, HAP-B, HAP-C permutation P = 0.00005). The finding of allelic association with markers in the UHMK1 gene might help explain why it has not been possible, despite great effort, to satisfactorily confirm previously reported associations between schizophrenia and the genes RGS4 and NOS1AP/CAPON. These genes flank UHMK1 and all three loci are within a 700 kb region showing linkage to schizophrenia. The confirmation of association between UHMK1 and schizophrenia, rather than RGS4 and NOS1AP in the London sample, points to the possibility that previous efforts to accurately fine map a gene in the 1q23.3 region have lacked accuracy or may have suffered from methodological flaws.