In a population of 49 male volunteers, mean daily alcohol consumption was measured first by a questionnaire and then twice by a diary. The reliability of the diary method was estimated to be at least 90%. The mean daily consumption reported in the questionnaire was on the average 60% of that revealed by the diary. The difference between the two methods in measuring alcohol consumption did not depend either on age, the level of alcohol consumption, the percentage of drinking days or on the percentage of days when the subjects felt intoxicated by alcohol. This shows that the ranking of subjects according to questionnaire data is not biased in a group of subjects motivated to give accurate answers.