Estimating premorbid intelligence in schizophrenia is difficult because the illness affects aspects of premorbid and postmorbid functioning. We evaluated two qualitatively different estimates of premorbid intelligence in a sample of schizophrenia patients and tested whether: (1) the two indices were related and produced similar IQ estimates, and (2) either index was related to a measure of cognitive deterioration. The Barona Index (BI, a demographically-based instrument) and the National Adult Reading Test (NART, a reading test of irregularly-spelled words) were utilized. Subjects (n = 40) were adult neuroleptic-medicated inpatients with a DSM-III-R diagnosis of chronic schizophrenia (n = 35) or schizoaffective disorder (n = 5). Paired t-tests revealed statistically equivalent BI and NART estimates for Full Scale and Verbal IQs, but significantly higher NART Performance IQs (t[35] = -3.34, p < 0.01). Correlational analyses suggested the two indices were associated but shared modest variance. BI correlations revealed expected associations with education and social position. NART IQs were related to education and a measure of cognitive status. Regression analyses supported the association between NART estimates and cognitive deterioration. Results suggest BI may be a better estimate of premorbid intelligence in schizophrenia as it is less influenced by potential consequences of the disease.