The validity of the Abbreviated Mental Test was demonstrated by comparison with final clinical diagnoses in a consecutive sample of 168 patients admitted with acute illness to a department of health care of the elderly. Fifty-eight (34%) had abnormal cognition. The best cut-off point was 8, with less than 8 suggesting abnormal cognitive function. A short version (the AMT7) of the AMT was developed. Its validity, internal consistency and coverage of domains was equivalent to the AMT but it had a slightly higher sensitivity (with acceptable specificity) than the original. This new short version may improve performance of junior doctors in clinical practice who appear to have difficulty remembering all 10 items of the AMT.