Consecutive series involving 172 patients with small cell lung cancer were analyzed retrospectively using eight pretreatment and two treatment-related prognostic factors in respect of their influence on survival. All the patients received chemotherapy with or without chest irradiation, according to phase II or phase III trial protocols of the National Cancer Center Hospital, Tokyo, from 1970 to 1987. The influence on survival of the various factors was investigated using univariate methods and Cox's proportional hazards model. In patients who survived for more than one year, a performance status of 0-1, limited disease, an age greater than 60 years, the absence of liver metastasis, radiotherapy to primary site and response to chemotherapy were determined by univariate analysis to be favorable prognostic factors. By multivariate analysis, performance status (P = 0.005), age (P = 0.026) and response to induction chemotherapy (P = 0.0001) proved to be valuable prognostic survival factors. The extent of disease which had been considered one of the most significant prognostic factors, was shown not to be a significant independent variable by multivariate analysis. Staging procedures may influence the prognostic analysis. Although classifying the limited or extensive stage of disease is still recommended, the current staging system lacks stringency and may not, in fact, reflect the tumor burden accurately. A simpler and meaningful staging system needs to adopted universally in order to continue the build-up of data for comparison from all institutions.