Approximately one-third of non-small cell lung cancer (NSCLC) patients present with locally advanced disease. Increasing numbers of clinical trials are being conducted in this group of patients and recently a new international staging system has been introduced, resulting in the sub-division of Stage III into IIIa (potentially operable disease) and IIIb (inoperable disease). Kaplan-Meier survival analyses and Cox regression analyses were used to analyze data from 129 Stage III NSCLC patients who had been treated on two consecutive Phase II trials testing combined modality treatment. The pretreatment characteristics of these patients were: median age--59 years, males/females--87/42, caucasian/non-caucasian--111/18, squamous cell or adenocarcinoma/large cell carcinoma--57/72, previous weight loss less than or equal to 5%/greater than 5%-76/46, previous history of cardiorespiratory disease--no/yes--91/36, performance status (PS) 0-1/2-3--102/27, Stage III, 2 groups--IIIa/IIIb--83/46, Stage III, 3 groups--IIIa T3 N0/IIIa N2/IIIb--41/41/47, surgical eligibility--eligible/ineligible--83/46. Kaplan-Meier statistics revealed significantly longer survival for PS 0-1 versus 2-3 (p = .001), for eligible versus ineligible for surgery (p = .003), for Stage-IIIa versus IIIb (p = .004), and for Stage-IIIa T3N0 versus IIIa N2 versus IIIb (p = .004). The best model developed from Cox regression analyses included stage (IIIa T3 N0 vs IIIa N2 vs IIIb), PS, and sex. These observations appear to have implications for clinical research in Stage III NSCLC.