Some experimental studies suggested that one possible oestrogen-receptor-unrelated mechanism of action of tamoxifen involves inhibition of angiogenesis. We evaluated the correlation of the degree of vascularisation of the primary tumour and we assessed it by using the panendothelial marker anti-CD31 and immunohistochemistry with microvessels count, performed at the vascular 'hot spot' of each single cancer, with the risk of recurrence in time. A cohort of 176 consecutive patients with node-positive invasive breast cancer treated with adjuvant tamoxifen (30 mg/daily for 3 years) and a median follow-up of 72 months was studied. Sixty-two patients developed metastasis (30 visceral, 18 skeletal and 14 in soft tissues) during the time of observation. The study of the hazard function for metastasis was performed by a generalized linear modelling approach with a binomial error according to Efron. The risk of first recurrence was strictly associated with vascular index, having the patients with the highest microvessel counts the highest risk of metastasis during all the period of observation. We did not find an interaction of vascularity with oestrogen receptor (ER) status. However, in the subgroup of patients with ER-positive tumours the hazard of metastasis was almost constant in time, while in that with ER-negative tumours it increased rapidly up to 20 months and, thereafter, decreased sharply. The results of our study are an indirect evidence that the patients with highly vascularized breast cancers may gain poor benefit of adjuvant tamoxifen and, therefore. that this antioestrogen is unlikely to retain a clinically relevant antiangiogenic activity in human breast cancer. Our data need confirmation by a prospective randomized clinical trial.