The Maudsley Personality Inventory (MPI) and its successor the Eysenck Personality Inventory (EPI) are frequently used to assess the personalities of psychiatric patients, and Eysenck advocates their routine administration as an aid to both diagnosis and treatment. However, a fundamental requirement of any personality test given to patients is that it should be relatively independent of the patient's mental state and reflect his personality rather than his illness. In depressive illnesses at least, it is doubtful whether this requirement is met. Coppen and Metcalfe (1965) administered the MPI to 39 patients with severe depression while they were depressed and again after recovery. Their mean N (neuroticism) score fell from 30.5 to 18.7 on recovery, and their mean E (extraversion) score rose from 17.3 to 20.6. Both changes were statistically significant. Assuming that the true scores were those obtained after recovery, it follows that the mean N score increased 63 per cent. and the E score decreased 19 per cent. during the episode of depression. The customary definition of personality as the relatively stable and enduring aspects of an individual's behaviour precludes such gross changes from being interpreted as changes in personality. One can only conclude, therefore, either that N and E scores obtained in the presence of severe depression are meaningless, or that they are a reflection of something other than the individual's personality. In the case of Coppen and Metcalfe's findings the most obvious explanation is that although patients are enjoined in the test instructions to rate their “usual way of acting or feeling” they do not in practice do so, but remain preoccupied with their current state of mind. Several MPI questions like “Does your mind often wander while you are trying to concentrate?” and “Have you often lost sleep over your worries?” are clearly relevant to the mental state of depression and probably attract positive replies from most patients regardless of their normal feelings.