The purpose of this pilot study was (1) to determine the proportion of patients in an ambulatory medical clinic who have an alcohol history recorded when prescribed benzodiazepines, and (2) to assess the adequacy of the alcohol history when obtained. Medical records of 35 outpatients who obtained prescriptions for benzodiazepines at a large inner-city teaching hospital medical clinic were audited. In none of the records was there evidence that the physician had sufficient knowledge of the patient's alcohol use to safely prescribe a benzodiazepine. In 57% of the records, no information about alcohol use was recorded. In the remaining 15 medical records, the information recorded was limited. The implications of prescribing benzodiazepines without knowledge of drinking status are discussed.