Twenty-four nude mice bearing MCF-7 breast cancer cells grown as xenografts and treated with tamoxifen (2.5 mg slow-release pellet) were studied for up to 35 days. Tumour size was measured in 2 dimensions at regular time-intervals and tumours were harvested on each of days 2, 4, 7, 14, 28 and 35 after the start of treatment. Control animals (8) received no treatment and the tumours were harvested after 0 or 35 days. Tumour sections were assessed for prevalence of apoptosis and mitosis and examined immunocytochemically for Ki(67)(MIB-1) and bcl-2 expression. Tumours increased in size during tamoxifen-treatment, but at a significantly slower rate (max. 2.6-fold) than in the untreated control animals; thus tumours not actually regressing may, nevertheless, be responding significantly to tamoxifen. MIB-1 and bcl-2 immunostaining and mitosis failed to show any consistent change over the period of study. Apoptosis, however, increased progressively and significantly to day-28 in tamoxifen-treated tumours, reaching approximately a 5-fold increase over day-0 values, then decreasing again to nearly 3-fold by day-35 (P= 0.0002). The apoptosis: mitosis ratio in treated tumours also increased to approximately 10-fold on day-28 over day-0 values, decreasing to nearly 4-fold by day-35 (P= 0.037). Within the treated group, apoptosis was significantly inversely correlated with both mitosis (R = -0.38, P= 0.03) and expression of bcl-2 (R = -0.48, P= 0.0056) and strongly positively correlated with both time on tamoxifen (R = +0.63, P= 0.0003) and the % inhibition of growth by tamoxifen (R = +0.58,P = 0.0012) in the 28 individual, treated tumours (estimated relative to the mean growth rate in the controls). The apoptosis: mitosis ratio was also inversely correlated with bcl-2 expression (R = -0.56, P= 0.0021) and positively correlated with both time on tamoxifen (R = +0.50, P= 0.0068) and % inhibition of growth (R = +0.56, P= 0.0019). In this hormone-sensitive tumour model for breast cancer, in which tamoxifen caused inhibition rather than regression, it was not possible to detect significant changes in the marker proteins Ki(67)and bcl-2, or in the prevalence of mitosis in relation to treatment; these factors may therefore not be accurate indices of response to tamoxifen in all situations. By contrast, however, tamoxifen induced a significant, early increase in the prevalence of apoptosis associated with inhibition of tumour growth and an inverse relationship in both mitosis and bcl-2 expression, suggesting that apoptosis may be an accurate and sensitive early marker of even a moderate response to tamoxifen.