After stratification for extent of small cell lung cancer, 109 patients were randomized to receive cycles of chemotherapy with cyclophosphamide, doxorubicin, and VP-16-213 [CAVP16 (regimen I)] or to receive CAVP16 to maximum response (minimum of three courses) and then chemotherapy with CCNU, methotrexate, vincristine, and procarbazine (COMP) alternating with CAVP16 (regimen II). A group of patients who achieved complete remission were randomized to receive whole-brain irradiation or to have observation only. Of the 44 patients with limited disease, 28 (64%) achieved a complete remission and 11 (26%) achieved a partial remission. Of the 65 patients with extensive disease, 26 (40%) achieved a complete remission and 28 (46%) achieved a partial remission. There were no significant differences between the regimens in response or survival. The projected median survival times are 14 and 10 months for limited and extensive disease, respectively. Nearly 30% of patients with limited disease will be 2-year, disease-free survivors. Twenty-nine patients were randomized to receive cranial irradiation or observation only; none of the 15 irradiated patients developed cerebral metastases, but five of 14 randomized to observation relapsed in the brain (P = 0.02). One patient died with necropsy evidence of only intracranial disease. The principal hematologic toxic effect was leukopenia. There were 31 febrile episodes (21 infectious) during neutropenia and four toxic deaths. Nonhematologic toxicity was mild. Cranial irradiation in patients who achieve complete remission delays or reduces the incidence of CNS metastases. Although alternating chemotherapy is not beneficial, combination chemotherapy with CAVP16 alone is highly effective treatment modality for small cell.