Prognostic evaluation of advanced liver disease is usually made on the basis of the common clinical and biochemical data included in the Child-Turcotte classification. The aim of this study was to evaluate the contribution of data from splanchnic angiography as a guide to prognosis in patients with cirrhosis. Over an 8-year period 219 patients with cirrhosis were investigated by splanchnic angiography and followed up prospectively. At the end of the study 95 patients had died (43.4%). Median survival time was 68 months. In addition to several clinical and biochemical data, hepatic portal venous perfusion and the presence of caudad hepatofugal veins as assessed by angiography were significant predictors of survival. Incorporating all nonangiographic variables in a Cox's multiple regression analysis, a clinicobiochemical set of prognostic covariates (ascites, s-albumin, gammaglobulins, s-alkaline phosphatase, and sex) was selected. When adding to this model each of the angiographic variables, only portal perfusion resulted in an independent predictor of survival. In conclusion, in cirrhotics the angiographic evaluation of portal perfusion improved the prognostic information obtained from clinical and biochemical data.