In a hospital-based case-control study of ovarian cancer conducted in Athens (1989-1991), 189 women with histologically confirmed common malignant epithelial tumors of the ovary were compared with 200 hospital visitor controls. All interviews were conducted by personal interview in the 2 participating hospitals and the data were analyzed by modelling through logistic regression, controlling for demographic and reproductive variables. Tranquilizing and hypnotic drugs (mostly diazepam) were not associated with risk of ovarian cancer: the adjusted relative risk (RR) and 95% confidence interval (CI) were 0.96 (0.57 to 1.63), whereas use of analgesics (mostly salicylates) was associated with significantly reduced risk of the disease (RR 0.51; CI 0.26 to 1.02). There was no evidence that perineal application of talc was associated with increased risk (RR 1.05; CI 0.28 to 3.98) but the frequency of reporting talc use was low in the study population. There was a statistically significant (p for trend 0.007) and a dose-dependent association between hair dyeing and risk of ovarian cancer. Compared to never-users, women dyeing their hair up to 4 times per year had a relative risk of 1.74 (0.91 to 3.32) whereas those dyeing their hair 5 or more times per year had a relative risk of 2.16 (1.19 to 3.89).