Twenty-four evaluable patients with stage T4 breast cancer were entered into a phase II study and received chemotherapy comprising cyclophosphamide 1,000 mg m-2 i.v., doxorubicin 50 mg m-2 i.v., vincristine 1.4 mg m-2 i.v. and prednisolone 40 mg orally for 5 days, given 3 weekly for four cycles prior to undergoing loco-regional radiotherapy. All patients completed treatment as planned with no major acute toxicity from either chemotherapy or radiotherapy. Subsequently 52 patients with stage T4 breast cancer were randomised in a phase III trial to receive either radiotherapy alone (RT) or this chemotherapy and radiotherapy (CHOP + RT). A significantly higher complete response rate was achieved in the CHOP + RT treatment arm (P = 0.03). However a larger proportion of the RT arm achieved loco-regional control after salvage treatment for relapse such that 50% of the RT arm and 57% of the CHOP + RT arm had no evidence of loco-regional disease at the time of last follow-up or death. There was no statistical difference in time to distant relapse or overall survival. Analysis of the pilot study showed results comparable to the trial CHOP + RT arm. This trial suggests that this cytotoxic therapy used in conjunction with radiotherapy has only marginal value in improving prognosis in locally advanced breast cancer.