The present retrospective study was undertaken to verify whether the extent of intratumour proliferative activity variation or the method of quantifying tumour proliferative activity is related to biological characteristics and clinical outcome in a series of operable node-negative breast cancer patients. For tumour proliferative activity evaluation, the 3H-thymidine autoradiographic assay was used. After incubation of 3-8 samples from different areas of the equatorial section of each tumour for 1 h at 37 degrees C with 3H-thymidine, the following methods were used for evaluation of tumour cell labelling: mean tumour labelling index (LI), the highest labelling value from a specific area (LI-max), and the extent of intratumour labelling variation from several samples (LI-CV). LI-max was related to ER and PgR status, and linearly correlated with LI (c.c. = 0.92, P < 10(-6)) whereas LI-CV was independent of tumour size, grade ER and PgR status, but dependent on the number of tumour samples analysed for each tumour. After 5 years of median follow-up, disease-free survival was only related to tumour size (T1 versus T2: 84 versus 64%, P < 0.04 by log rank analysis) and different LI values (low versus high 3H-Tdr-LI:86 versus 61%, P < 0.03 by log rank analysis). LI-max and LI-CV values were not significantly related to clinical outcome. Cox multivariate analysis confirmed the independent prognostic value of LI and tumour size on disease-free survival.