Genetic variation in CYP17 is suspected to be related to endometrial cancer risk based on its role in the regulation of steroid and non-steroid hormone biosynthesis. Reported associations between CYP17 and higher levels of estradiol in some studies suggest that the C allele of a T-to-C single nucleotide polymorphism (SNP) in the 5'UTR of CYP17 (rs743572) may be associated with an increased risk of hormone-related cancers. However, five relatively small epidemiologic studies of endometrial cancer have reported that women with the rs743572 C allele have a decreased risk of endometrial cancer. To examine this association, we genotyped rs743572 and eight other haplotype-tagging SNPs (htSNPs), which are estimated to capture >80% of the variation in CYP17 in a population-based study of 497 endometrial cancer cases and 1,024 controls in Poland. Significant associations were not found for rs743572 (per C allele: OR = 1.12, 95%CI 0.96-1.30; P-trend = 0.15), for individual htSNPs, or for extended haplotypes (global P-value = 0.60). When we pooled data from previously published studies with our own (a total of 1,004 endometrial cases and 1,907 controls), we observed significant study heterogeneity in summary estimates of the association between rs743572 and endometrial cancer, as well as evidence of publication bias. In conclusion, our data are not consistent with a decreased endometrial cancer risk associated with rs743572, as previously reported, or with other haplotype-tagging polymorphisms. Further evaluation in consortia is necessary to confirm potential weak associations between common variation in CYP17 and endometrial cancer risk and to address the concern of publication bias.