Tests of word-reading ability, such as the North American Adult Reading Test (NAART) and Wide Range Achievement Test (WRAT) Word Reading subtest, are often used by neuropsychologists to estimate premorbid intelligence. However, despite the fundamental basis in reading, little work has been done to calculate or estimate literacy level from NAART performance. Additionally, few prior studies have evaluated the use of word-reading tests in underrepresented populations such as African Americans or individuals with lower socioeconomic status and educational attainment. Finally, while the correlations between the NAART and older editions of the WRAT are known, no study has yet examined the relationship between the NAART and WRAT-Fourth Edition (WRAT-4). The purpose of the current study, then, was to replicate and extend prior research in these areas. NAART and WRAT-4 Word Reading (WRAT4WR) performances were examined in a sample of 173 primarily African American individuals who were HIV-positive and who had an average of fewer than 12 years of education. Analyses revealed a strong correlation between NAART error scores and WRAT4WR raw scores (r = -.866, p < .001). Linear regression was used to establish the predictive equation to estimate WRAT4WR scores from NAART errors. This information allows for determining reading grade-level equivalents based on NAART error scores.