In 103 patients with small-cell lung cancer, we compared four courses of standard doses of Adriamycin (A) (Adria Laboratories, Columbus, Ohio), vincristine (V), and cyclophosphamide (C) with a regimen of increased doses of cyclophosphamide and to a lesser extent, Adriamycin. We found no significant difference in rate (22% v 21%) or median duration (seven v nine months) of complete remission. Patients not in complete remission after the four cycles of AVC received two courses of VP-16 (etoposide) and cisplatin: the complete remission rate increased to 49% and 48% respectively. Patients on the high-dose arm received co-trimoxazole prophylaxis; those on the standard arm did not. Patients on the high-dose arm had a higher incidence of neutropenia (nadir less than 500 cells/microL) but a lower incidence of infection for similar degrees of neutropenia. However, they also suffered more severe side effects of a different kind. Cotrimoxazole thus allowed for the administration of higher doses of chemotherapy to outpatients by protecting them from infection. However, the higher doses of cyclophosphamide and Adriamycin, did not improve treatment results, produced more severe side effects, and is not recommended.