The suggested link between angiogenesis in breast cancer and metastasis remains unsubstantiated. We tested this relationship in primary breast carcinomas from 37 patients with a median follow-up 9.5 years (Cohort 1) and 50 patients with a median follow-up of 1.5 years (Cohort 2). Angiogenesis was assessed by counting vessel density after immunohistochemical staining of vascular endothelium for factor VIII. Patients were grouped according to whether metastasis (defined as spread to axillary lymph nodes, distant sites or both) had occurred. The mean +/- SD scores in Cohort 1 when metastasis was absent and present, respectively, were 15.6 +/- 4.9 (n = 21) and 14.1 +/- 3.7 (n = 16). In Cohort 2 the scores were 15.4 +/- 5.8 (n = 26) and 14.5 +/- 4.9 (n = 24). There was no significant difference between these scores in either cohort. Multivariate analysis demonstrated lymph node involvement (P < 0.001) and tumour size (P < 0.001) but not angiogenesis score (P > 0.05) to predict distant metastasis. This evidence argues against any prognostic significance of angiogenesis in breast carcinoma.