The relationship between hormone replacement treatment (HRT) and breast cancer risk was analysed using data from a case-control study conducted between June 1991 and February 1994 in six Italian centres on 2569 patients aged below 75 with histologically confirmed breast cancer and 2588 controls admitted to hospital for a wide spectrum of acute, non-neoplastic, non hormone-related diseases. Ever HRT use was reported by 7.5% of cases and 7.5% of controls, corresponding to a multivariate odds ratio (OR) of 1.2 [95% confidence interval (CI), 0.9-1.5]. The risk increased with increasing duration of use: the ORs were 1.0 for use lasting less than 1 year, 1.3 for 1-4 years and 1.5 for 5 years or more. There was no clear pattern of risk with reference to time since starting use, but the OR was significantly elevated (OR = 2.0, 95% CI 1.3-2.9) for women who had stopped HRT within the last 10 years. No association was observed in those who had stopped HRT more than 10 years ago (OR = 1.0). The increased OR for women who had stopped HRT within the last 10 years was consistent across strata of identified covariates, and was significantly related to duration of use. This study confirms the absence of a strong association between HRT and breast cancer risk, although the risk estimate was above unity for women who had used HRT for 5 years or longer. However, the risk was significantly elevated in the short to medium term after use, particularly for long-term use. This short-term increased risk is consistent with an effect of HRT on one of the later stages of the process of breast carcinogenesis. The flattening of risk with increasing time since stopping, and hence the absence of a long-term cumulative excess in breast cancer risk after stopping HRT exposure, has relevant implications on individual risk assessment and public health.