AIMS: To examine (a) effects of different modes of data collection on the reporting of alcohol consumption and non-response rate, and (b) differences in reported consumption between respondents and non-respondents. DESIGN: Two versions of a health questionnaire survey were assigned to two random samples, one version to each sample. Version 1 consisted of a telephone interview without alcohol questions, followed by a mailed questionnaire with alcohol questions. Version 2 consisted of a telephone interview with alcohol questions, followed by a mailed questionnaire without alcohol questions. SETTING: Participants were recruited randomly in eight Swiss cantons. PARTICIPANTS: Five hundred and thirty-seven (404) respondents to the telephone interview (and subsequent mailed questionnaire) with version 1, and 451 (360) with version 2. MEASUREMENTS: Alcohol-use variables derived from a quantity-frequency measure. RESULTS: Respondents to the mailed questionnaire (version 2) did not differ significantly in alcohol consumption from non-respondents. Response rate was not affected by inclusion of alcohol questions, but respondents asked by telephone about their alcohol use were more often abstainers and less often hazardous drinkers than respondents to the mailed question. CONCLUSION: The study gives no indication that interviews are refused because alcohol consumption is a questionnaire topic, but suggests that postal questionnaires give slightly greater disclosure of alcohol consumption.