Seventy-eight patients with cirrhosis were prospectively followed for up to 20 months, on the average. At entry into the study, galactose elimination capacity, aminopyrine breath test, and ICG clearance were measured. At the end of the study, 27 patients had died. Univariate analysis using the Kaplan-Meier method showed that both quantitative liver function tests (galactose elimination capacity: P less than 0.025; aminopyrine breath test: P less than 0.001; ICG clearance: P less than 0.005) and common clinical and biochemical data (encephalopathy: P less than 0.001; ascites: P less than 0.001; serum bilirubin: P less than 0.005; serum albumin: P less than 0.001; prothrombin index: P less than 0.05) were significant predictors of survival. To investigate whether quantitative liver function tests could contribute to a better definition of the prognosis, once Pugh score had already been taken into account, a multiple regression analysis according to the Cox model was performed. Pugh score and galactose elimination capacity resulted in the only independent prognostic covariates. From them a prognostic index was calculated, and the model was validated in an additional sample of 70 patients investigated according to the same protocol. The contribution GEC gave to the assessment of overall prognosis over that obtained using the Pugh score was slight, as estimated by the statistical parameters of the Cox's model, but was significant as assessed by a ROC curve analysis (P = 0.05). These data show that all quantitative liver function tests were predictors of survival in cirrhosis, and that the galactose elimination capacity added some new prognostic information to those already available using the Child-Turcotte-Pugh classification.