Treatment choice in primary breast cancer is wide and still controversial; it seems likely that the optimum treatment for individual patients could be dictated by biological indicators of tumour behaviour. If biopsy could provide prognostic information as well as detailed tissue diagnosis then definitive treatment, with or without adjuvant systemic therapy, could be planned from the outset. We studied 140 patients with a clinical diagnosis of primary breast cancer to determine how much information could be obtained from Tru-Cut needle biopsies performed at the first clinic visit. Ten patients were found to have benign disease. Of 130 carcinomas, 123 (95 per cent sensitivity) were diagnosed correctly from the needle biopsies, with seven false negative and no false positive results (100 per cent specificity). Precise histopathology was predicted in 121 (93 per cent). Grade was correctly assessed in 77 of 112 (69 per cent), but needle biopsy was not accurate for assessment of lymphatic invasion nor elastosis. Steroid hormone receptors were assayed in 59 needle biopsies, and the incidence of oestrogen receptor positivity (34, 58 per cent) was similar to the resected tumours (35, 59 per cent), but the incidence of progesterone receptor positivity (26, 44 per cent) was lower (33, 56 per cent, P less than 0.04). Immunostaining with monoclonal antibody human milk fat globule membrane was accurate in the needle biopsies. DNA analysis by flow cytometry was performed in 37 tumours and the concordance between needle biopsies and resected samples was high. Tru-Cut needle biopsy obviates open biopsy and gives reliable detailed information.