Neoadjuvant chemotherapy for large operable breast cancers is being increasingly used for the purpose of "downstaging," so that the lesions become accessible to conservative treatment. Since tumor proliferative activity has been shown to be a major prognostic factor for breast cancers, we have studied the value of S-phase fractions established by flow cytometry on cytological samples at diagnosis, in 184 stage II or IIIa patients entered in a randomized trial comparing neoadjuvant to adjuvant chemotherapy. All patients were pre- or perimenopausal, and the median follow-up was for 43 months (24-64). Using the median value (5%) as cutoff, a high SPF was found to be associated with relapse (p < 0.0008), locoregional recurrence (p < 0.02), or metastasis (p < 0.003). However, when the patients were analyzed according to the type of treatment, significance was maintained for the patients in the primary radiotherapy arm (p < 0.003) but not for those in the neoadjuvant chemotherapy arm (p < 0.06). The overall rate of response to neoadjuvant chemotherapy was significantly lower for tumors with low SPF (56.5%), compared to tumors with high SPF (85.6%). Thus, SPF was no longer predictive of outcome when the tumors regressed by more than 50% after chemotherapy (p = 0.66), whereas it was highly predictive in the nonresponding patients (p < 0.0001). Our study has revealed that patients with low-SPF tumors, irrespective of response or treatment schedule, had similar prognosis (around 70% free of disease at 45 months), while the high-SPF nonresponders had a dismal prognosis, with less than 25% free of disease at 24 months. If our results are confirmed with a longer follow-up, proliferative activity of breast cancers should prove to be instrumental for the initial therapeutic decision of stage II or IIIa patients.