Polymorphic variants of microsomal epoxide hydrolase (mEPHX) with altered enzyme activity have been associated with an increased risk for ovarian cancer. We assessed the frequency of exon 3 and exon 4 variants of mEPHX among 291 ovarian cancers and 257 controls from a UK-based population. The distribution of the exon 3 alleles among both the cancer and control groups was significantly different from that expected under Hardy-Weinberg equilibrium suggesting that the PCR restriction fragment length polymorphism (PCR-RFLP) genotyping assay might be flawed. The codon 113 polymorphism was reassessed using a two-color allele-specific PCR-based assay. We found that a codon 119 G>A polymorphism, present in 20% of the British population and linked to the wild-type exon 3 allele, resulted in some Tyr113/His113 heterozygotes being falsely classified as His113/His113 homozygotes when using the PCR-RFLP assay. Consequently, we reassessed all our codon 113 data using the new allele-specific assay. We found no evidence of an association of ovarian cancer risk with the exon 3 Tyr113>His113 variant. Similarly the frequencies of the exon 4 His139>Arg139 genotypes were not significantly different between cases and controls. Stratifying the genotyping data according to the predicted mEPHX activity revealed a highly significant decrease in high mEPHX activity among the serous ovarian cancers (P=0.01) suggesting that high mEPHX activity may be protective for this histological sub-type. Furthermore previous disease association studies of exon 3 alleles which utilized the PCR-RFLP assay may be compromised by the existence of a codon 119 G>A polymorphism which may be common in Caucasian populations.