Although the effect of malnutrition on survival has been demonstrated by a number of studies, it is not clear whether malnutrition represents an independent risk factor in patients with liver disease. We studied 212 hospitalized patients with liver cirrhosis who were followed clinically for 2 y or until death. Body fat and muscle mass were evaluated by triceps skinfold thickness (TSF) and midarm muscle circumference (MAMC), respectively. Multivariate analysis according to Cox's model assessed the predictive power of nutritional parameters on survival. Thirty-four percent of patients had severe malnutrition as determined by MAMC and/or TSF below the 5th percentile and 20% had moderate malnutrition (MAMC and/or TSF < 10th percentile). Twenty-six percent of patients were overnourished (MAMC and/or TSF > 75th percentile). Severely and moderately malnourished patients had lower survival rates than normal and overnourished patients. When analyzed with Cox's regression analysis, severe depletion of muscle mass and body fat were found to be independent predictors of survival. The inclusion of MAMC and TSF in the Child-Pugh score, the prognostic score used most with liver disease, improved its prognostic accuracy. The prognostic power of MAMC was higher than that of TSF. These data demonstrate that malnutrition is an independent predictor of survival in patients with liver cirrhosis. The inclusion of anthropometric measures in the assessment of these patients might provide better prognostic information.