The prognostic value of breast cancer proliferative activity was evaluated in 385 women operated for primary, non-metastasised mammary carcinoma. Cell kinetics was measured using two immunohistochemical techniques. Cells in S-phase of cell cycle were labelled in vitro by incubation of fresh tissue fragments with 5-bromo 2-deoxyuridine (BrdU), a thymidine analogue. Nuclei of cells in active DNA synthesis were stained by an anti-BrdU monoclonal antibody (Mab). Cells in interphase and mitosis were detected with Ki-67, a Mab that is known to react with a nuclear antigen present in G1/S/G2/M phases of cell cycle, but not in resting cells. This reagent provides a means of evaluating the growth fraction of neoplastic cells. BrdU was incorporated in a proportion of tumour cells ranging from 0.1 to 65.5% (median 6.8%). In the panel of tumours presented in this report the median percentage of Ki-67 positive cells (Ki-67 score) was 9.0% (range 0.1-77%). The relationship between disease-free survival (DFS), BrdU labelling index, Ki-67 score and 13 different clinico-pathological variables was investigated by multivariate analysis, using the Cox proportional hazards model. Axillary node status (P = 0.009) and Ki-67 score (P = 0.038) emerged as independent prognostic factors. Nodal status and tumour growth fraction allowed division of patients into groups at different risk of relapse: tumours with a proliferative index below the median value showed a lower recurrence rate than tumours with a high proliferative activity (P < 0.001). In particular, no relapse occurred in pN0 patients bearing carcinomas with a Ki-67 labelling < 9.0% 4 years after surgery. These findings suggest that the evaluation of proliferative activity in breast cancer enhances the probability of correctly predicting outcome after surgery and could be of assistance in the planning of adjuvant therapies.