Cancer and Leukemia Group B (CALGB) accrued 1,745 patients with limited (LD) or extensive (ED) small-cell lung cancer (SCCL) to five separate trials between 1972 and 1986. We reviewed these data to evaluate the impact of pretreatment prognostic factors on outcome. In multivariate analysis, female gender was predictive of improved response (LD, P = .01; ED, P = .04) and survival (LD, P = .01; ED, P = .02). A performance status of 0 or 1 was associated with improved response rates in both subsets, but was statistically significant (P = .04) only for overall objective response in LD patients. Performance status was a highly significant predictor of survival in both LD and ED groups (P less than .001). Supraclavicular lymph node involvement, while still LD, had a borderline unfavorable impact on survival (P = .06) compared with a lesser extent of LD involvement. In ED patients, a decrease in survival rates was associated with an increased number of metastatic sites (P = .01). Changes in the patient population were noted with time: the percentage of women increased from 21% to greater than 35%; an increased number of metastatic sites was identified among ED patients; mean performance status improved for both LD and ED subsets. These trends reflect the changing demographics of lung cancer, improved lung cancer staging, and probably lead-time bias. Response rates, overall survival, and long-term (greater than 2-year) survival varied significantly among the five protocols, both before and after multivariate correction for identified prognostic variables. However, the changing character of the study population limits the ability to determine retrospectively how much improvements in therapy contributed to the positive changes in failure-free survival, overall survival, and long-term survival observed in our sequentially studied population.