On the basis of the glutamatergic dysfunction hypothesis of schizophrenia, we have been conducting a systematic study of the association of glutamate receptor genes with schizophrenia. Here we report association studies of schizophrenia with polymorphisms in three kainate receptor genes: GRIK3, GRIK4 and GRIK5. We selected 16, 24 and 5 common single nucleotide polymorphisms (SNPs) distributed in the entire gene regions of GRIK3 (>240 kb), GRIK4 (>430 kb) and GRIK5 (>90 kb), respectively. We tested associations of the polymorphisms with schizophrenia using 100 Japanese case-control pairs (the Kyushu set). We observed no significant "single marker" associations with the disease in any of the 45 SNPs tested except for one (rs3767092) in GRIK3 showing a nominal level of significance. The significant association, however, disappeared after the application of the Bonferroni correction. We also observed significant haplotype associations in seven SNP pairs in GRIK3 and in four SNP pairs in GRIK4. None, however, remained significant after Bonferroni correction. We also failed to replicate the nominally significant haplotype associations in a second sample set, the Aichi set (106 cases and 100 controls). We conclude that SNPs in the gene regions of GRIK3, GRIK4 or GRIK5 do not play a major role in schizophrenia pathogenesis in the Japanese population.