Estrogen replacement therapy (ERT), in common use among menopausal women, has been inconsistently related to risk of ovarian cancer. Part of the inconsistency may be due to differences in risk according to tumor histology. We therefore compared, within histologic groups, 367 cases of invasive epithelial ovarian cancer to 564 population controls on history of ERT practices. In multivariate logistic regression analyses adjusted for age, parity, oral-contraceptive usage, and other factors, the relative odds of nonmucinous ovarian cancer increased by 5.1% for each year of ERT use [i.e., adjusted odds ratio, 1.051; 95% confidence interval (CI), 1.01-1.09, as a multiplicative factor for each year of use]. In particular, compared to never use, women who employed ERT for a total of 5 years or longer had a relative odds of 2.03 (95% CI, 1.04-3.97) of having a serous carcinoma and 2.81 (95% CI, 1.15-6.89) of having an endometrioid carcinoma. Mucinous tumors were not associated with use. ERT usage may contribute to the development of nonmucinous types of ovarian cancer, and additional histology-specific analysis of this association in other studies is warranted.