One hundred fifty-eight evaluable patients with Stage II carcinoma of the breast with positive lymph nodes were treated either with adjuvant chemotherapy or irradiation therapy followed by chemotherapy. Patients were randomized to test the effectiveness of L-phenylalanine mustard (L-PAM) with and without postoperative irradiation therapy (R.T.) against cyclophosphamide, 5-fluorouracil, and methotrexate (CMF) with and without postoperative irradiation therapy in decreasing the frequency of recurrence. No significant difference was observed between L-PAM as compared with R.T. plus L-PAM (P = 0.85). The difference between CMF and R.T. plus CMF was significant in support of CMF alone (P = 0.04). The frequency of recurrence between R.T. plus L-PAM and R.T. plus CMF showed no difference. A comparison of the chemotherapy only regimens showed an advantage of CMF over L-PAM (P = 0.05). Both the delay in starting chemotherapy and the significant decrease in percent optimal drug dosage during the first six cycles of therapy are factors that may influence the high frequency of relapse observed in the R.T. plus chemotherapy groups. Of the four treatments, CMF has the lowest frequency of recurrence.