In a series of 137 patients with cancer of the vulva who had undergone radical surgery we investigated, by means of a Cox regression model, which combination of clinical variables and clinical as well as histopathological variables afforded the best prediction of survival/death from cancer and survival/death from other causes. Among clinical variables the best prediction was afforded by a combination of tumor site in the clitoris/not clitoris, tumor size greater or smaller than 40 mm, obesity/not obesity, and age. Among clinical and histopathological variables the best prediction was by a combination of groin node metastases/not groin node metastases, tumor site in the clitoris/not clitoris, degree of differentiation high/moderate + low, and age. The combination of clinical and histopathological variables had a significantly better predictive power than clinical variables alone. The best prognostic group in both combinations had a 5-year-survival for cancer of 98%, while the poorest prognostic group in the two combinations showed a 5-year survival for cancer of 19 and 9%. The best predictors of death from causes other than cancer were age and a poor general health. It is concluded that the poorest prognostic group is definitely under treated and that the best prognostic group is presumably over treated. Patients in poor general health and with a good cancer prognosis should receive a more conservative treatment.