Since the estrogen receptor alpha (ER) is an important mediator of hormonal responses such as proliferation in estrogen-sensitive tissues, we hypothesized that polymorphisms in the ER gene could be functional and associated with endometrial cancer risk. We performed a population-based case-control study in Sweden, focusing on restriction fragment length polymorphisms for XbaI and PvuII and an upstream TA repeat polymorphism. In the main analysis, 154 cases and 205 controls who never used hormone replacement therapy took part and we calculated age-adjusted and multivariate odds ratios (OR) and 95% confidence intervals (CI) using unconditional logistic regression. The XbaI X allele appeared to confer a reduced risk for endometrial cancer. The multivariate OR for the XX genotype was 0.52 (95% CI 0.21-1.29) compared to the xx genotype and there were suggestions of decreasing risk with increasing number of X alleles (P for trend = 0.07). The PvuII PP genotype was also associated with a non-significantly decreased risk for endometrial cancer (multivariate OR 0.70, 95% CI 0.34-1.44) compared with the pp genotype (P for trend = 0.43). The multivariate OR for two short TA (<19 repeats) alleles versus two long alleles was 1.54 (95% CI 0. 73-3.27) and there were suggestions of increasing risk with increasing number of short alleles (P for trend = 0.26). We observed the same pattern of results in an expanded group of subjects, which included women who had used hormone replacement (in total 288 cases and 392 controls). Our data suggest that variants of the ER gene may be associated with an altered risk of endometrial cancer.