PURPOSE: To predict the long-term survival results of clinical trials earlier than using actuarial methods and to assess the factors predictive of long-term cure in patients with limited-stage small-cell lung cancer. METHODS AND MATERIALS: Between 1981 and 1998, 1417 new cases of small-cell lung cancer were diagnosed in Saskatchewan, Canada, of which 244 were limited stage and treated with curative intent. They were followed to the end of February 2002. A parametric lognormal statistical model was retrospectively validated to determine whether long-term survival rates could be estimated several years earlier than is possible using the standard life-table actuarial method. RESULTS: The survival time of the uncured group followed a lognormal distribution. Four 2-year periods of diagnosis were combined, and patients were followed as a cohort for an additional 2 years. The estimated 10-year cause-specific survival rate was 13% by the lognormal model. The Kaplan-Meier calculation for 10-year cause-specific survival rate was 15% +/- 3%. The data also showed that the absence of mediastinal lymphadenopathy and higher chest radiotherapy dose were significant prognostic factors on multivariate analysis (p < 0.05). Among the 163 patients given prophylactic cranial irradiation, a higher biologically effective dose to the brain did not improve survival or decrease the incidence of brain metastases. CONCLUSION: The lognormal model has been validated for the estimation of survival in patients with limited-stage small-cell lung cancer. A higher biologically effective dose to the brain did not improve survival or decrease the incidence of brain metastases.