The current study examines the relationship between problem alcohol use and severity of liver disease by self-administered questionnaires using both the CAGE questionnaire and beverage-specific quantity-frequency questions. The cohort consisted of 38 patients with cirrhosis (10 with decompensated liver disease) and 62 with mild fibrosis (stage 0–1), of comparable mean age and estimated duration of infection. Although mean alcohol consumption was similar in both groups, a positive CAGE screen (defined as two or more affirmative answers) was significantly more common among cirrhotics (OR = 5.24; 95% CI, 1.78–15.39) and tended to be associated with decompensated liver disease (OR = 13.3; 95% CI, 0.67–256) among cirrhotics. In multivariate analysis, only inflammatory grade on liver biopsy (OR = 67.7; 95% CI, 10.6–431) and positive CAGE score (OR = 8.09; 95% CI, 1.15–57.1) were independent predictors of cirrhosis. These findings suggest that the CAGE questionnaire predicts advanced liver disease better than daily or lifetime drinking measures.