There have been many randomised trials of adjuvant tamoxifen among women with early breast cancer, and an updated overview of their results is presented. In 1995, information was sought on each woman in any randomised trial that began before 1990 of adjuvant tamoxifen versus no tamoxifen before recurrence. Information was obtained and analysed centrally on each of 37000 women in 55 such trials, comprising about 87% of the worldwide evidence. Compared with the previous such overview, this approximately doubles the amount of evidence from trials of about 5 years of tamoxifen and, taking all trials together, on events occurring more than 5 years after randomisation. Nearly 8000 of the women had a low, or zero, level of the oestrogen-receptor protein (ER) measured in their primary tumour. Among them, the overall effects of tamoxifen appeared to be small, and subsequent analyses of recurrence and total mortality are restricted to the remaining women (18000 with ER-positive tumours, plus nearly 12000 more with untested tumours, of which an estimated 8000 would have been ER-positive). For trials of 1 year, 2 years, and about 5 years of adjuvant tamoxifen, the proportional recurrence reductions produced among these 30000 women during about 10 years of follow-up were 21% (SD 3), 29% (SD 2), and 47% (SD 3), respectively, with a highly significant trend towards greater effect with longer treatment (chi2(1)=52.0, 2p<0.00001). The corresponding proportional mortality reductions were 12% (SD 3), 17% (SD 3), and 26% (SD 4), respectively, and again the test for trend was significant (chi2(1) = 8.8, 2p=0.003). The absolute improvement in recurrence was greater during the first 5 years, whereas the improvement in survival grew steadily larger throughout the first 10 years. The proportional mortality reductions were similar for women with node-positive and node-negative disease, but the absolute mortality reductions were greater in node-positive women. In the trials of about 5 years of adjuvant tamoxifen the absolute improvements in 10-year survival were 10.9% (SD 2.5) for node-positive (61.4% vs 50.5% survival, 2p<0.00001) and 5.6% (SD 1.3) for node-negative (78.9% vs 73.3% survival, 2p<0.00001). These benefits appeared to be largely irrespective of age, menopausal status, daily tamoxifen dose (which was generally 20 mg), and of whether chemotherapy had been given to both groups. In terms of other outcomes among all women studied (ie, including those with "ER-poor" tumours), the proportional reductions in contralateral breast cancer were 13% (SD 13), 26% (SD 9), and 47% (SD 9) in the trials of 1, 2, or about 5 years of adjuvant tamoxifen. The incidence of endometrial cancer was approximately doubled in trials of 1 or 2 years of tamoxifen and approximately quadrupled in trials of 5 years of tamoxifen (although the number of cases was small and these ratios were not significantly different from each other). The absolute decrease in contralateral breast cancer was about twice as large as the absolute increase in the incidence of endometrial cancer. Tamoxifen had no apparent effect on the incidence of colorectal cancer or, after exclusion of deaths from breast or endometrial cancer, on any of the other main categories of cause of death (total nearly 2000 such deaths; overall relative risk 0.99 [SD 0.05]). For women with tumours that have been reliably shown to be ER-negative, adjuvant tamoxifen remains a matter for research. However, some years of adjuvant tamoxifen treatment substantially improves the 10-year survival of women with ER-positive tumours and of women whose tumours are of unknown ER status, with the proportional reductions in breast cancer recurrence and in mortality appearing to be largely unaffected by other patient characteristics or treatments.