To determine whether the National Adult Reading Test (NART) would provide a valid estimate of premorbid intelligence in schizophrenia, two schizophrenic samples were recruited, one consisting of 35 patients resident in long-stay wards, the other of 29 patients normally resident in the community. Schizophrenic patients were individually matched for age, sex, and education with a healthy, normal subject. Both schizophrenic samples scored significantly lower on the Wechsler Adult Intelligence Scale (WAIS) than their respective control groups. NART-estimated IQ did not differ significantly between the community-resident schizophrenics and their controls, suggesting that the NART provides a valid means of estimating premorbid intelligence in such a population. NART-estimated IQ was significantly lower in the long-stay sample than in their controls. Although low NART scores in this latter sample could be a valid reflection of low premorbid IQ, the alternative explanation that NART performance was impaired by onset of the disease cannot be ruled out.