The purpose of the present paper was to evaluate correlations between clinical response to chemotherapy and outcome in a subgroup analysis of premenopausal patients with tumours considered too large for breast conserving surgery, treated with primary chemotherapy (n = 200) from a previously published trial (Scholl S.M., Fourquet A., Asselain B, et al. Eur J Cancer 1994, 30A, 645-652). Objective response rates amounted to 65% following four courses. In a multivariate Cox regression analysis, comparing seven parameters, the following variables were associated with poor survival: clinically involved nodes [N1b:RR: 2.7 (95% CI 1.3-5.3)], the failure to respond to chemotherapy [D:RR: 2.62 (95% CI 1.3-5)] and a raised S phase fraction [SPF > 5%: RR: 2.4 (95% CI 1.2-5)]. Parameters associated with increased metastatic recurrence rates, by order of entry in the model, were: young age [< 35: RR: 2.46 (95% CI 1.2-5)], large clinical tumour size [T3: RR: 2.02 (95% CI 1.2-3.4)], poor histological grade (SBR III: RR: 1.93 (95% CI 1.1-3.3)] and the failure to respond to chemotherapy [D: RR: 1.91 (95% CI 1-3.4)]. The assessment of both tumour cell proliferation rates as well as possibly drug resistance markers (although not available in the present study) should be helpful in selecting patients likely to benefit from intensified chemotherapy regimens. The most accurate predictor of response in the present study appeared to be the response to chemotherapy treatment itself.