The National Adult Reading Test (NART: Nelson, 1982) has become the standard means of estimating premorbid intelligence. The danger in using the NART for this purpose is that it yields an invalid estimate if a client's performance on the test has suffered impairment. In the present study a sample of 659 healthy subjects was used to build a regression equation for the prediction of NART scores from demographic variables (i.e. years of education, social class, age and sex). The multiple correlation between these demographic variables and the NART was .70 (p < .0001). Comparing a client's obtained and predicted NART score will permit the clinician to assess objectively whether NART performance is impaired, and thus whether or not the NART will provide a valid estimate of premorbid intelligence. 1990 The British Psychological Society
