In order to test an alcoholism screening strategy in a primary care population of a polyclinic, we administered two standardized questionnaires, the Michigan Alcoholism Screening Test (MAST) and the Cut-down, Annoyed, Guilty, Eye-opener Test (CAGE), to all consecutive patients seen during a week. The applicability rate was 77% (269/350). Reasons for being inapplicable were inability to understand the test questions or restricted time allowed for each visit. Tests were positive in 19 and 22%, for the MAST and CAGE, respectively. Kappa coefficient of agreement between tests was 0.69. Based on a positive MAST (cutoff > or = 5), alcoholism was more frequent in males than in females (27 vs 10%), in patients aged 40-49 years (males 45%, females 29%), in unemployed than in employed people (32 vs 22%), in native citizens than in foreigners (21 vs 17%). 23% of MAST positive patients had one or more alcohol-related somatic diseases. Only 60% of MAST positive patients were known to physicians as alcoholics or alcohol-abusers. In conclusion, we showed that (1) prevalence of alcoholism is high in this primary care population, (2) agreement between MAST and CAGE is good, and (3) alcoholism screening is better performed by standardized questionnaires than the usual clinical investigations.