Among patients with resected non-small cell lung carcinoma, about 50% will present a tumor recurrence. Thus, it would be of major importance to be able to predict and try to prevent these relapses by an active chemotherapy and/or radiotherapy. In an attempt to answer this question, the tumors of 227 patients with a surgically resected non-small cell lung carcinoma were evaluated as follows: tumors were classified as squamous cell carcinoma (n = 132) or adenocarcinoma (n = 95), and tumor differentiation was evaluated for each type. Then, all tumors were classified in respect to their pathological TNM staging (WHO) and screened by immunohistochemistry for the detection of the expression of the following antigens: Bcl-2, A+B+H blood group antigens, c-erb-b2, p53, and Pan-Ras antigens. Furthermore, adenocarcinomas were screened for the presence of point mutations in Ki-Ras codons 1-31. Finally, the patient blood group was defined, and patient survival was analyzed using nonparametric tests and proportional hazard Cox models. Using Kaplan-Meier survival curves, disease pathological TNM staging was shown to be a strong predictive factor of survival for both squamous cell carcinoma and adenocarcinoma. Patients with squamous cell carcinoma experienced fewer relapses than those with adenocarcinoma (42% versus 63%; P = 0.0002) and had a significantly better survival. All evaluated antigens were more often present in squamous cell carcinoma than in adenocarcinoma except for Pan-Ras (three times more frequent in adenocarcinoma). In patients with squamous cell carcinoma, only tumor staging had a significant prognosis value (P = 0.01). In patients with lung adenocarcinoma, a well-differentiated tumor (P = 0.009) as well as a positive Bcl-2 staining (P = 0.009) and an A+B+H antigen tumor staining (P = 0.024) were associated with a better survival. In contrast, patients with a stage I or II disease and a p53-positive tumor staining and patients with the O blood group (P = 0.01) had a shorter survival. Interestingly, no relation with patient survival was related to c-erb-b2 and Pan-Ras staining. Finally, 12 point mutations were found out of 81 tumors (15%) evaluated for Ki-Ras codons 1-31; they involved codon 12 but also 8, 14, and 15 without any relationship to survival. In respect to lung adenocarcinoma, using Cox proportional hazard models stratified on tumor staging, the following markers were shown to be related to survival: (a) Independent markers of longer survival (ie., high histological degree of tumor differentiation and positive Bcl-2 and A+B+H blood group antigen expression by tumor cells); and (b) Independent markers of shorter survival (i.e., O blood group for all patients and p53 tumor staining in patients with stage I and II diseases). This study suggests that, in patients who undergo surgery for lung adenocarcinoma, the presence or absence of these criteria could be used to define a subset of patients who may benefit from a more specific follow-up.