Information provided by patients about the amounts of alcohol they drink may often be too subjective and therefore unreliable. Because of the possible serious consequences of interactions between alcohol and medication, reliable laboratory test markers for alcohol consumption are needed. Carbohydrate-deficient transferrin (CDT) is at present the best available objective measure of drinking behavior. During a withdrawal trial, 92 alcohol-dependent patients who had been admitted to a hospital in an ethanol-intoxicated state were monitored over the following 28 days by using the percent carbohydrate-deficient transferrin (%CDT of total transferrin) (%CDT) method. At the time of admission, 63% showed elevated %CDT levels. After a subsequent period of abstinence, a decrease in %CDT levels was apparent in four different groups of patients, whereas in two groups, comprising the greatest number of patients, normal %CDT levels were evident after 14 days of abstinence. In patients whose CDT levels were very high at study initiation, it took at least 21 to 28 days--and sometimes longer--for CDT to decrease to the radioimmunoassay (RIA) %CDT test cutoff point of 2.5. In a further study of 56 male alcohol-dependent patients, we measured liver enzyme concentrations, mean corpuscular volume (MCV), and four CDT variants on the first day of evidence of withdrawal syndrome. We found a significant correlation between results on the Munich Alcoholism Test (MALT) and MCV levels; among gamma-glutamyltransferase (GGT), aspartate aminotransferase (AST), and alanine aminotransferase (ALT) levels; and among all four CDT variants. A cluster analysis yielded three clusters: (1) GGT, AST, and ALT levels; (2) MCV levels and MALT results; and (3) all CDT measurement variants. We conclude that these three clusters measure different detriments to the patient and that all available CDT variants are commensurate.