This paper reports the preliminary results of a cohort study of 4544 British women receiving hormone replacement therapy (HRT). These women were recruited at 21 specialist menopause clinics around Britain. Up to the end of June 1983, the mean duration of HRT use per woman was 67 months of which, on average, 43% was 'opposed' use. In general, however, both the amount of progestogen given and the number of days per cycle for which it was given was less than would have been the case if the women had been receiving modern opposed therapy. The major focus of the study was to monitor mortality and cancer incidence in the cohort. The mortality results were broadly reassuring: overall mortality was significantly lower than expected on the basis of national rates (relative risk 0.58) and mortality ratios were below unity for all specific causes of death examined apart from cancer of the ovary (relative risk 1.43, 95% confidence limits 0.62-2.82) and suicide or suspected suicide (relative risk 2.53, 95% confidence limits 1.26-4.54). The most likely explanation for the latter finding is selection bias--thus at least 7 of the 11 women who died from suicide or suspected suicide had a psychiatric history before receiving HRT. The cancer incidence results were less reassuring, although they should be interpreted with some caution because cancer registry rates were used for comparative purposes. With this proviso, endometrial cancer risk was significantly elevated (relative risk 2.84, 95% confidence limits 1.46-4.96); many of the women concerned had taken therapy which was predominantly or entirely opposed although only one woman had received an opposed regimen which would now be considered adequately protective to the endometrium. Breast cancer incidence was also significantly increased (relative risk 1.59, 95% confidence limits 1.18-2.10); detailed analysis suggested that the use of unopposed ethinyl-oestradiol in particular might have undesirable effects on the breast.