In order to detect and quantify intellectual impairment current IQ test performance must be compared with an estimate of premorbid intellectual level. One means of obtaining this estimate is to use a current ability measure which meets, as fully as possible, the following three criteria; firstly, the measure should be reliable, secondly, test performance should correlate highly with IQ in the general population, thirdly, test performance should be resistant to the effects of neurological and psychiatric disorder. In the present work, the National Adult Reading Test (NART; Nelson, 1982) was evaluated against these criteria. The NART is a single word, oral reading test and consists of 50 words of irregular pronunciation (eg ache, deny). The results indicated that the NART has high split-half, test-retest and inter-rater reliability. Secondly, the NART has high construct validity as a measure of intelligence (ie it loads highly on g) and predicts a substantial proportion of WAIS IQ variance. Thirdly, NART performance was largely resistant to the effects of neurological and psychiatric disorder. In this last respect it was superior to the Vocabulary subtest of the WAIS, a commonly used, putative premorbid index. An alternative method of estimating premorbid IQ was developed by Wilson et al. (1978) in the US. Using the WAIS standardisation sample, these authors built a multiple regression equation to estimate premorbid WAIS IQ from demographic variables (eg education, occupation). In the present work, demographic regression equations were built to estimate WAIS IQ in the UK population. The results indicated that the demographic approach is applicable in the UK. Demographic variables predicted 50&'37 of WAIS IQ variance. It was also demonstrated that regression equations which combine the NART and demographic variables provide more accurate estimates than either method alone. While the demographic approach is a less powerful predictor of IQ than either the NART or combined method, its compensatory advantage is that the estimates it provides are entirely independent of current cognitive capacity. The clinical utility of premorbid indices was demonstrated by the fact that they improved discrimination between impaired and healthy subjects over the use of current IQ measures alone. To date, UK research on premorbid indices has used WAIS IQ as the criterion variable. The final aim of the present work was to build equations to estimate premorbid WAIS-R IQ. Drawing on the results of earlier work, a revised version of the NART was developed (NART-R). Equations for the estimation of WAIS-R IQ from the NART-R, demographic variables and the combination of these variables are presented.