Liver transplantation for alcoholic cirrhosis remains controversial. In particular, criteria for the selection of patients who will remain recovered from alcoholism post-transplant require better definition. We analyzed the long-term predictive value of categorizing transplant referral patients with alcoholism and end-stage liver disease into risk groups for recidivism and noncompliance. Forty-seven patients with the diagnosis of alcoholism and advanced liver disease were evaluated and placed into predefined risk groups (low-, moderate-, and high-risk) for recidivism and noncompliance. No absolute period of abstinence from alcohol was required. All patients were asked to sign a contract not to drink alcohol and comply with a rehabilitation program before and after transplantation. Compliance with alcohol rehabilitation, abstinence, functional level, employment, and survival were assessed. Patients who were not compliant with the rehabilitation program or consumed alcohol were scored as failures. Thirty-one patients were ranked as low risk, and were accepted for liver transplantation; 27 patients were transplanted. Five of 31 patients (16%) drank alcohol. One patient drank before and four patients drank transiently after transplantation. Ten patients were categorized as moderate risk, and were deferred for transplantation; two patients underwent later transplantation. All 10 patients (100%) were noncompliant or drank alcohol, including two patients who drank after transplantation after a period of abstinence and rehabilitation. Six patients were ranked as high risk, and were denied liver transplantation. Five patients (83%) drank alcohol and were noncompliant. Minimum follow-up was 12 months (mean, 24 months; range, 12-41 months). The mean Karnofsky performance score was 34 before and 84 after liver transplantation. Actuarial survival of alcoholic patients undergoing transplantation was 93%. We conclude that categorization of transplant referral patients with alcoholism and liver failure into predefined risk groups for recidivism and noncompliance accurately predicts pre- and post-transplant behavior. As defined, only low-risk alcoholic patients are good candidates for liver transplantation.