During the last decade, the Eastern Cooperative Oncology Group (ECOG) has studied a series of combination chemotherapy regimens in metastatic (stage IV) non-small-cell lung cancer (NSCLC). In January 1984, the ECOG activated a randomized study, EST 1583, which concluded the evaluation of combination regimens in phase III trials and initiated the evaluation of single agents exclusively in previously untreated patients. The treatment regimens in EST 1583 consisted of: (1) mitomycin, vinblastine, and cisplatin (MVP); (2) vinblastine and cisplatin (VP); (3) MVP alternating with the regimen cyclophosphamide, doxorubicin, methotrexate, and procarbazine (CAMP); (4) carboplatin followed by the MVP regimen at the time of progression; and (5) iproplatin followed by MVP at the time of progression. From January 1984 to July 1985, 743 patients were entered on this trial and 699 fulfilled the eligibility requirements. The following objective response rates (complete plus partial remissions) were observed: first-line MVP, 20%; VP, 13%; MVP/CAMP, 13%; carboplatin, 9%; iproplatin, 6%; and second-line MVP, 6%. First-line MVP produced a significantly higher response rate than the other treatments (P = .03) adjusted for prognostic variables. Using analyses that were adjusted for prognostic covariates, survival for patients treated on a given regimen was compared with survival for all remaining patients. These analyses showed that treatment with carboplatin was associated with longer survival (median survival time, 31.7 weeks; P = .008) while initial treatment with MVP was associated with a trend for shorter survival (median survival time, 22.7 weeks; P = .09). It should be noted that none of these regimens appear to have produced a clinically meaningful prolongation of survival. Similar analyses evaluating time to progression disclosed that carboplatin-treated patients had a significantly longer time to progression (median time to progression, 29 weeks) than all remaining patients (P = .01). Life-threatening and lethal toxicities (toxicity grades 4 and 5) were greater on the combination regimens than on the single agents (P less than .0001). Based on these results, current group-wide ECOG trials in stage IV NSCLC consist of randomized phase II trials evaluating single agents.