OBJECTIVE To assess the usefulness of the biochemical marker 'carbohydrate deficient transferrin' (CDT) in relation to conventional markers for chronic excessive alcohol use. METHODS Prospective. METHODS Addiction clinic Paschalis, Wanssum, the Netherlands. METHODS Addicts for weaning (n = 125) were questioned at admission about their drinking habits in the last two weeks. Based on the criterion more or less than 60 g alcohol per day, the group was divided into excessive and nonexcessive alcohol users (men: 52 abusers, 51 non-abusers; women: 12 abusers, 10 non-abusers). Mean cell volume (MCV), gamma-glutamyl transpeptidase (gamma GT) and total transferrin were measured in blood collected 2 days after admission, as well as CDT by two methods (CDTect and % CDTriTIA). RESULTS In men the CDTect test was the most sensitive: sensitivity 82% with specificity 88%. The sensitivity and specificity were 62% and 86% for gamma GT, 50% and 95% for % CDTriTIA, and 34% and 98% for MCV. The combination of a positive CDTect result and a positive gamma GT result gave a predictive value of use of alcohol > 60 g/day of 100%. The results of CDT and gamma GT were also used for a logistic regression model, giving a statistical prediction for excessive alcohol use. The subgroups of women were too small to detect statistical significant differences between tests. CONCLUSIONS The CDTect test was more sensitive for the detection of chronic excessive alcohol use than the conventional markers. The combination of gamma GT and CDTect results increased the positive predictive value.