The purpose of this study was to develop a sensitive as well as brief screening questionnaire by combining the well-known instruments CAGE and the Michigan Alcoholism Screening Test (MAST) in detecting patients with alcohol dependence or abuse in general hospitals and general practices. The number of items was reduced by means of logistic regression and item analysis based on data of 1,167 consecutive admissions of a general hospital who completed both questionnaires. Further data were derived from a sample of 774 patients from 10 randomly selected general practices. A solution with nine items was validated in a second sample of 436 hospital inpatients. In all three samples, cases screening positive were interviewed using the Schedules for Clinical Assessment in Neuropsychiatry to provide ICD-10 and DSM-III-R or DSM-IV diagnosis. In addition, 103 subjects with negative screening results were interviewed in the second general hospital sample. On grounds of the data of all three samples, a solution of seven items was chosen. This instrument comprises two CAGE and five MAST questions (Leubeck Alcohol Dependence and Abuse Screening Test; LAST) and was significantly higher in sensitivity than CAGE and SMAST. Data were robust in all three samples. It is concluded that the LAST is an optimized instrument for use in general hospital and general practice.