The relationship between hormonal therapy for menopause (hormone replacement therapy, HRT) and the risk of epithelial ovarian cancer was evaluated in a collaborative re-analysis of 4 European case-control studies, 2 conducted in Greece and 1 each in Italy and the United Kingdom, including a total of 1,470 ovarian cancer cases and 3,271 hospital controls. Odds ratios (ORs) for HRT use were derived after allowance for study centre, age, socio-economic level, parity, menopausal status, type of menopause, age at menopause and oral contraceptive use. Overall, 109 (8.0%) ovarian cancer cases and 146 (4.7%) controls had ever used HRT, corresponding to an adjusted OR of 1.71 (95% confidence interval 1.30-2.25). The point estimates of the OR were 1.77 in the first Greek study, 1.40 in the second Greek study, 1.66 in the Italian study and 1.68 in the British study. Adjustment for possible confounders, including menopausal status, type of menopause, age at menopause and oral contraceptive use, slightly increased the OR. Limiting the analysis to women with information on relevant aspects of HRT use revealed a weak positive association with duration and some evidence that the excess relative risk for ovarian cancer declined with time since last use. These findings are compatible with a promoting effect of HRT in ovarian carcinogenesis. It is also possible, however, that the positive association reflects chance or selective administration of HRT to high-risk individuals, since until recently in Europe HRT was prescribed mainly for alleviation of peri-menopausal symptoms.