This study examined the utility of word recognition scores for estimating actual Verbal IQ scores obtained from 1–5 years earlier. Participants were 271 persons remaining normal and 24 initially normal persons who developed cognitive impairment over longitudinal follow-up. A previously published regression equation based on education and the American modification of the National Adult Reading Test (Nelson, 1982) was used to estimate Mayo Verbal IQ. In spite of correlating well with prior obtained scores (r = .7) the predicted score tended to overestimate the obtained Mayo Verbal IQ. A revised equation was developed in the normal sample, which improved accuracy of prediction. Among the 24 persons who developed cognitive impairment over a 5-year span, the revised predicted scores provided reasonable estimates of initial Mayo Verbal IQ. To aid in clinical interpretation, a table of the normal frequencies of predicted Mayo Verbal IQ minus contemporaneously obtained Mayo Verbal IQ is provided. (JINS, 1997, 3, 528–533.)