A randomized clinical trial was performed to determine if the addition of hormonal therapy with tamoxifen to a combination chemotherapy regimen was superior to the chemotherapy alone for adjuvant treatment of premenopausal women after mastectomy for node-positive breast cancer. The chemotherapy regimen utilized consisted of cyclophosphamide (C), 5-fluorouracil (F), and prednisone (P), and the doses employed were: C, 150 mg/m2 IV days 1 to 5; F, 300 mg/m2 IV days 1 to 5; and P, 10 mg orally three times daily on days 1 to 7. A total of ten courses of therapy, given every 6 weeks, was planned. Tamoxifen (T) was given at a dose of 10 mg twice daily and was stopped 6 weeks after the last course of CFP. Four hundred patients are fully eligible and evaluable. With a median observation time of 5.3 years, the proportion of recurrences on each arm were: CFP, 95 of 202 (47%); CFPT, 77 of 198 (39%). The relapse-free survival distribution for CFPT was superior to that for CFP, at a borderline level of significance (two-sided P = 0.06). When significant prognostic factors were considered in covariate analysis, CFPT was not significantly better than CFP (P = 0.43). This marked change in level was due to imbalance in several factors not considered in stratification. Currently, 31% of CFP and 25% of CFPT patients have died, and although there is a slight separation of the survival curves in favor of CFPT, the difference is not significant (P = 0.21). Analysis within receptor subsets also showed no significant advantage for the addition of tamoxifen. This study does not establish a significant advantage for the concurrent administration of tamoxifen with the CFP regimen. It does, however, clearly demonstrate the importance of examination of clinically important prognostic factors, even those not utilized in stratification, and consideration of these factors in covariate analysis if imbalances are present.