Two hundred and fifty patients were entered into a randomized clinical study to compare the effectiveness of goserelin (Zoladex) in depot formulation with diethyl stilboestrol in locally advanced or metastatic prostate cancer. In 22 patients from the two arms of the study regular assessments were made of the effect of these hormone treatments on the haemostatic system. Selection of those patients with no recent surgical intervention and those on no drugs liable to interfere with the haemostatic mechanism was done at entry, in order to remove bias and achieve comparable groups. Baseline comparison of the two treatment groups showed no difference in clinical or biochemical measures of disease extent or activity, including serum prostate specific antigen (PSA) levels. There was a significant fall in plasma antithrombin-III (AT-III) activity in the DES treated group both from baseline and compared with the goserelin group. This effect commenced within 1 month and was maintained until monitoring ceased at 12 months. There was also a significant increase of fibrinolytic activity in the DES treated patients compared with those on goserelin. No divergence between the two treatment groups was seen in any other haematological parameters at baseline or on follow-up. A single AT-III estimation was also performed on a larger group of 74 patients at median follow-up time of 17 months (range 3-24). This confirmed the difference noted in the original study group. In the main study thrombotic episodes were noted in 13/126 patients treated with DES and 0/124 treated with goserelin (P less than 0.001). These findings suggest that lowered AT-III is the major factor through which DES affects the coagulation mechanism, and that no such effect is seen with goserelin treatment despite an equivalent therapeutic efficacy.