SummaryThe relationship between female hormone use and primary liver cancer was analysed using data from a case-control study conducted between 1984 and 1992 in Milan on 82 female incident cases with histologically or serologically confirmed hepatocellular carcinoma and 368 controls admitted to hospital for acute non-neoplastic, non-hormone-related diseases. An elevated relative risk (RR) or primary liver cancer was observed in oral contraceptive (OC) users (RR 2.6, for ever versus never users, 95% confidence interval, CI 1.0-7.0). The RR was directly related to duration of use (RR 1.5 for A5 years and 3.9 for >5 years) and persisted for longer than 10 years after stopping use (RR 4.3%, 95% CI 1.0-18.2). The RR were below unity, although not significantly, for women ever using oestrogen replacement therapy (RR 0.2, 95% CI 0.03-1.5) and female hormones for indications other than contraception and menopausal therapy (RR 0.4, 95% CI 0.1 -1.5). The long-lasting, association between risk of hepatocellular carcinoma and OC use has potential implications on a public health scale, since primary liver cancer is a relatively rare disease among young women, but much more common at older ages. This study provides limited but reassuring evidence on the possible relationship between oestrogen replacement treatment and subsequent risk of hepatocellular carcinoma.There is evidence from several case-control studies of a positive association between the use of oral contraceptives (OC) and primary liver cancer in developed countries (Henderson et al., 1983;Neuberger et al., 1986;La Vecchia et al., 1989;Palmer et al., 1989;Yu et al., 1991;Hsing et al., 1992;Trichopoulos, 1992; WHO, 1992), although data are inconsistent for developing countries (WHO, 1989). This association is biologically plausible, as OCs are promoters of hepatocarcinogenesis in rodents (Yager & Yager, 1980) and the pill is known to increase the risk of adenoma of the liver in humans (Baum et al., 1973;Mettlin & Natarajan, 1981).In Italy, mortality from primary liver cancer in women below age 45 is extremely low, although some upward trend between the 1950s and the 1980s has been observed in young women, but not in young men (Decarli & La Vecchia, 1984). This may support an involvement of OC use in the pathogenesis of this cancer. A case-control study found a significantly increased risk of hepatocellular carcinoma in Italian women 32-59, who had used the pill for longer than 5 years (La Vecchia et al., 1989 Relative risks (RR) of liver cancer and the corresponding 95% confidence intervals (CI) in relation to OC, oestrogen replacement treatment and female hormones for other indications were estimated, after adjustment for age, by the method described by Mantel and Haenszel (1959); for multiple levels of exposure, the significance of the linear trend in risk was assessed by the Mantel test (Breslow & Day, 1980). Unconditional multiple logistic regression, fitted by the method of maximum likelihood, was used to allow for possible confounding factors (Breslow & Day,...