This prospective study assessed the role of aminopyrine breath test in the prognosis of patients with cirrhosis, and evaluated whether the test provided useful information not included in the Pugh score. During a period of 36 months, 125 patients with biopsy proven liver cirrhosis were included, and followed for up to 48 months (median 17 months). During follow up 43 patients died (20 of liver failure). Survival was univariately related to aminopyrine breath test (p less than 0.02), Pugh score (p less than 0.01), presence of ascites (p less than 0.01), and sex (p less than 0.05). Using Cox's regression analysis, Pugh score, aminopyrine breath test, and sex, were independent significant predictors of survival. From the Cox's model a prognostic index was computed. According to a receiver operating characteristic curve analysis, the prognostic index predicting death showed an improvement in area under the curve when compared with a prognostic index calculated excluding aminopyrine breath test, but the improvement did not reach statistical significance (p = 0.12). A similar prognostic index was calculated to predict death from liver failure. Cox's regression analysis selected aminopyrine breath test, Pugh score, and aetiology as the best set of predictor covariates. According to a receiver operating characteristic curve analysis, a prognostic index cut off value of 2.6 had a 94% sensitivity and a 88% specificity. The prognostic index significantly improved prognostic accuracy when compared with a prognostic index calculated from Pugh score and aetiology, but excluding aminopyrine breath test (p = 0.05). These data disclose that the aminopyrine breath test offers additional prognostic information to the Pugh score, and the prognosis of patients with cirrhosis.