Between August 1983 and March 1985, a randomized study was conducted that compared cisplatin (CDDP) (80 mg/m2 on day 1) alone with CDDP plus vindesine (VDS) (3 mg/m2 on days 1, 8, and 15) in 160 consecutive patients with inoperable non-small cell lung cancer (NSCLC). There were no complete responses. The response rate for CDDP plus VDS (22 of 77 patients, 29%) was significantly higher than that for CDDP alone (9 of 78 patients, 12%) (P less than 0.05). However, no difference existed in the median duration of response (20 weeks for CDDP plus VDS versus 20 weeks for CDDP alone) or the median survival time (45 weeks for CDDP plus VDS versus 39 weeks for CDDP alone). No significant differences in toxicity were detected between the two arms; myelosuppression, alopecia, and peripheral neuropathy occurred more frequently with CDDP plus VDS and there was one lethal episode of hepatorenal syndrome in the CDDP plus VDS arm. Among the variables Eastern Cooperative Oncology Group (ECOG) performance status (PS), age, sex, stage, weight loss, serum lactate dehydrogenase (LDH) level, albumin level, histologic cell type, and chemotherapy arm, only chemotherapy arm was a significant factor leading to a major response (P = 0.019, multiple logistic regression analysis). The significant predictors of survival were PS (P = 0.000), sex (P = 0.000), and stage (P = 0.002) (Cox's proportional hazards model), with a PS of 0 or 1, female sex, and lower stage yielding the best survival. Although a significantly higher response rate was obtained in the combination arm than in the single agent arm, the survival benefit to patients receiving such combination chemotherapy was not determined and more effective chemotherapy regimens are required.