The authors studied the influence on survival of 21 clinical, anatomical, haematological and biochemical factors evaluated, at diagnosis, of 411 patients (pts) with advanced Non Small Cell Lung Cancer (NSCLC) followed in our department between 1984 and 1990. Most of the patients were male (347--84.4%) and only 64 (15.6%) were females. Median age was 62 years, but was slightly higher in females. Only 34 patients were aged under 45 years. Squamous cell carcinoma (215 pts--52%) and adenocarcinoma (152 pts--37%) were the most frequent histologic types. Performance status was poor--only 103 (25%) continued active; 120 (29%) spent at least half of the time in bed; 188 (46%) were severely limited. After staging, 179 (44%) presented locally advanced disease (stage IIIB) and 232 (56%) metastatic dissemination (stage IV). Therapy was defined by the oncologic group according to individual characteristics and based on clinical grounds. Anti-neoplastic therapy was performed in 225 (55%), chemotherapy alone in 121 (30%), radiation therapy alone in 67 (16%), and sequential combined treatment (chemotherapy and thoracic radiation) in 37 (9%). Until 1987, the main chemotherapy regimen was MACC (Metrotrexate + Adriamycine + Cyclophosphamide + Lomustin), afterwards VP(M) (Cisplatin + Vimblastin + Mitomycine). Radiation therapy was performed using Co60, 2 Gy/day, 5 days a week, for 4 weeks (approximately 45 Gy total). The response rate was poor--four complete responses (2%), 42 (19%) partial responses. The overall median survival was 4.3 months and only 5% of patients were alive after 18 months of follow up. Prognostic importance of each characteristic studied was initially done by unifactorial analysis, followed by multifactorial analysis according to two methods: Cox proportional hazards model and recursive partitioning amalgamation--RECPAM. Regardless of the method used, the main determinants of survival were found to be performance status (Zubrod), weight loss and serum albumin. Other factors such as the staging (presence or absence of metastasis), lymphocytes, lactic dehydrogenase, and hoarseness were also significant. It is noteworthy that age and histological type were irrelevant; sex and hoarseness only proved important when integrated within a multifactorial model. The overall prognostic evaluation and therapeutic decision of advanced NSCLC patients could be improved by combining the prognostic value of TNM with that of performance status, weight loss and serum albumin. These prognostic guidelines must be taken into account when designing new clinical trials.