A screening questionnaire was used to identify persons with alcohol-related problems in a general hospital and in a general practice. The questionnaire consisted of 16 questions about alcohol-related problems during the past 12 months, and two questions which assessed the quantity and frequency of usual alcohol consumption. The questions about alcohol-related problems led to the identification of more persons with problems than did the consumption questions alone. If the consumption questions only had been used to identify a high-risk group of consumers, and if all these subjects had been persuaded to reduce their alcohol consumption to a low-risk level, the reduction in the number of persons who experienced alcohol-related problems would have been modest. This is compatible with the phenomenon of the preventive paradox, and it is concluded that any screening questionnaire must include questions other than those about high-risk levels of consumption if it is to identify the majority of at-risk persons.