All consecutive eligible patients with non-small-lung carcinoma seen at Centro di Riferimento Oncologico and Istituto Nazionale per la Ricerca sul Cancro were entered into a randomized chemotherapy study. Conditions of eligibility included advanced stage (stage III not amenable to radiation therapy or i.v.), measurable or evaluable lesions, age less than 70 years, performance status (PS) greater than 40, and no previous chemotherapy. Patients were randomized to either CAMP (cyclophosphamide 300 mg/m2 i.v., adriamycin 20 mg/m2 i.v., methotrexate 15 mg/m2 i.v. days 1 and 8, procarbazine 100 mg/m2 orally from day 1 to day 10, every 4 weeks) or DE (cisplatin 20 mg/m2 i.v. for five consecutive days and etoposide 75 mg/m2 i.v. on the same days, every 3 weeks). Treatment was continued until progression. Out of the 136 patients randomized, 133 were eligible (CAMP 62, DE 71) and 108 evaluable. Patient characteristics included male/female ratio 57/5 (CAMP) and 61/10 (DE), median age of 60 years (CAMP) and 59 years (DE), PS greater than or equal to 70 for 39 (CAMP) and 50 (DE), PS less than 70 for 23 (CAMP) and 21 (DE), stage III for 18 (CAMP) and 15 (DE), and stage IV for 44 (CAMP) and 56 (DE). DE was superior to CAMP in terms of response rate, defined as responding/evaluable patient ratio (38.2% versus 20.8%); however, the responding/eligible patient ratio was not significantly different in the two groups. The superiority of DE tended to be more marked in stage III patients, in patients with PS greater than or equal to 70, and in the squamous histological type. Toxicity was acceptable (one toxic death) and evenly distributed in the two treatment groups; only renal toxicity was prevalent in the DE group. Survival (all eligible patients) was significantly better in the DE than in the CAMP group. Whether DE chemotherapy is superior to a no-chemotherapy approach has not been evaluated in this study and remains to be determined.