The objectives are to identify and integrate through regression analysis those fundamental clinical variables predicting survival of patients with inoperable lung cancer managed in a modern setting. Median survival time from first treatment in 129 patients with limited disease and 187 patients with extensive disease was 36 and 14 weeks, respectively. Within the proposed survival model for limited disease, weight loss was the major prognosticator followed by symptom status, supraclavicular metastases, and age. Within extensive disease, symptom status and age were dominant variables followed by weight loss and metastases to liver, opposite hemithorax, brain, and bone. Survival by cell type was similar within the limited and extensive disease groups. The data identify the essential factors which must be controlled or accounted for in studies analyzing survival as a dependent variable.