Serum levels of carbohydrate-deficient transferrin (CDT) were measured in subjects of two independent studies using two different commercial kits. The kits measure CDT either as a percentage of total transferrin (AXIS %CDT, AXIS Biochemicals AS, Norway), or as the absolute amount (CDTect, Pharmacia, Sweden). In a population of males (mean age 41 years) consisting of alcoholics, heavy, moderate and non-drinkers, a strong correlation was found between AXIS %CDT and CDTect results (r = 0.92, n = 58, P < 0.001). Sensitivity and specificity in detecting chronic alcoholic drinking of over 60 g/day were 78 and 94% for the AXIS assay, and 83 and 88% for the CDTect assay, respectively. In a population from a birth cohort study, consisting of 21-year-old males and females with less excessive alcohol consumption, the correlation between AXIS %CDT and CDTect CDT was weaker but still statistically significant (r = 0.46, n = 212, P < 0.001). In this population, with specificities > 83% in detecting alcohol consumption levels of > or = 6 drinks per week, the sensitivities were low with both CDT assays (< 43% for > or = 6 drinks per week, and < 44% for > or = 16 drinks per week). These results suggest that (a) both assays are equally effective in detecting chronic drinking over 60 g/day in older alcoholic males, and (b) both assays are similarly ineffective in detecting less excessive regular drinking in young males and females.