Eleven clinical and pathological factors were eligible for study in relation to the survival of 617 female patients with breast cancer treated by radical mastectomy. All the patients have been followed for more than five years. The data were analyzed by the Cox's proportional hazards regression model. A final set of five independent significant prognostic factors was obtained. In order of importance, they were clinical stage, age at diagnosis, axillary lymph node status, histologic type and size of the primary. To test the time-dependency of these factors using Cox's model, all except age reduced their prognostic impact significantly after five years, however, age was a powerful prognostic factor for long-term survival. A prognostic index based on the regression coefficients was constructed. This index can give a better prediction for our patients' survival and be taken as reference when formulating treatment protocols for breast cancer.