A multivariate prognostic index based on clinical data and the results of flow cytometry for the grading of breast cancer was evaluated in 117 patients whose disease had been detected and treated by mastectomy with axillary clearance between 1974 and 1976. Survival analysis with Cox's regression model pointed to three important prognostic factors: lymph node involvement (p less than 0.001), DNA ploidy (p less than 0.01) and tumour size (p less than 0.01). These factors were incorporated into a prognostic index, in which the lymph node involvement, DNA ploidy, and tumour size contributed to the index in that order. Logistic discriminant analysis with five year follow-up as the fixed end point (70 alive, 47 dead) gave the same result; lymph node involvement, tumour size, and DNA ploidy were the best prognostic indicators of survival. The result showed that our multivariate prognostic index was more powerful than lymph node involvement alone. The use of this prognostic index is recommended for selecting patients for different treatments.