A hospital-based case-control study of ovarian cancer was conducted in Athens from 1989 to 1991. The cases were 189 women, residents of Greater Athens and less than 75 years old, with histologically confirmed common malignant epithelial tumors of the ovary, operated in the 2 major cancer hospitals of the Greater Athens area. Controls were women residents of Greater Athens, less than 75 years old, who had never had cancer or had an ovary removed and who had visited patients hospitalized in the same wards as the ovarian cancer cases at the same time. The data were analyzed by modeling through multiple logistic regression. Statistically significant associations were found with induced menopause without oophorectomy [relative risk (RR) 0.17; 95% confidence interval (CI) 0.04 to 0.72], age at menopause (for an increment of 5 years RR 1.42; CI 1.00 to 2.01), hormone-replacement therapy (RR 5.73; CI 1.07 to 30.80), parity (RR 0.48; CI 0.24-0.96) and, marginally, age at first birth (for an increment of 5 years RR 1.30; CI 0.99 to 1.70). Non-significant but previously suggested or biologically plausible associations were noted with use of oral contraceptives (inverse), weight before onset of the disease (positive), and consumption of more than 2 glasses of alcoholic drinks per day (positive). There were no consistent associations with coffee, tobacco, moderate alcohol intake, broad occupational group, induced abortions, or age at menarche.