OBJECTIVE (1) Examine physician and nursing staff compliance with conducting an alcohol screening interview; (2) Compare compliance with the interview with usual physician and nurse assessment of drinking; (3) Examine reasons why drinking information might not be collected. METHODS Residents and nurses were taught how to use an alcohol screening interview and were told by the director of residency training or by the Vice-President for Nursing to administer it to all admitted patients. Data on interviewed patients were compared with medical record data on a randomly selected series of 80 patients who were not interviewed. METHODS Teaching hospital in an urban/suburban community. METHODS Residents on internal medicine and family practice services; surgical nurses. RESULTS Compliance with conducting the interview was low (14.7% of residents' admissions; 13.53% of nurses' admissions). Review of the medical records indicated that some alcohol-related information was recorded in most patients' medical records (physicians recorded information on 94% of patients on the teaching services, nurses on the surgical unit recorded alcohol-related information on 71% of patients). Surgeons and surgical residents recorded alcohol-related information on 30% of their patients. Residents were more likely to interview male than female patients, and residents and nurses tended to interview patients with higher GGTP values. CONCLUSIONS (1) Compliance with administering a standardized alcohol screening interview was low. (2) Nurses, and residents in family practice and internal medicine made some assessment of drinking for most patients. Surgeons assessed drinking in a minority of patients. (3) Expressed reasons for not administering the standardized interview included discomfort with the interview, viewing the interview as too time-consuming, or not part of the usual responsibilities of the nurses or residents. Other possible reasons for the low levels of compliance are discussed.