The morphology of the tumor and the anatomic extent of the disease are important factors influencing treatment selection and ultimately survival for patients with lung cancer. The American Joint Committee TNM system provides a method for consistent reproducible description of the primary tumor (T), the status of the regional lymph nodes (N), and the presence or absence of distant metastasis (M). The TNM subsets thus classified can be grouped into three "stages" of disease such that the survival expectations for patients in each stage and cell type are similar. This classification of patients with respect to estimates of their prognosis is essential for valid comparisons of treatment modalities and meaningful communication of end results information.Clinical characteristics which influence survival are reflected in the staging recommendations. The size of the lesion, the proximal margination, and the presence or absence of other pulmonary complications are features which distinguish the T classification as T1, T2, or T3. The presence or absence of lymph node involvement has an important bearing on survival expectations. Advancing from no nodal involvement, N0, to involvement of the peribronchial and hilar nodes, N1, and then to the mediastinal nodes, N2, causes progressive erosion in survival expectations. The tumor morphology and specific nodes that are involved are important components of this relationship. The presence of distant metastasis, M1, is synonymous with an extremely poor prognosis. Using these prognostic elements, the TNM subsets are combined into three stages of disease so that patients in each group will have a generally similar life expectancy, the survival for patients with stage I disease being significantly greater than that for patients with stage II disease which is significantly greater than survival for patients with stage III disease.Improvements in the outcome for lung cancer patients depend upon the depth and scope of our scientific understandings and our ability to communicate our observations to one another. Measures of response to treatment can be translated into therapeutic practice only if uniform evaluators are used. Accordingly, a reproducible valid system for staging of lung cancer is recommended.