A study was conducted to determine the validity of two separate item classification schemes for the reading comprehension subtest of the McGraw-Hill Test of Adult Basic Education, Level D, Form 4. The first of the two schemes was developed by the publisher and presented as a means of classifying errors for the purpose of diagnosing individual learning needs. The second was developed by the researcher through a subjective examination of the 45 subtest items. Data for the study consisted of test results from 242 adult students enrolled in an urban career assessment program. In the first stage of the study, a factor analysis was used to determine whether the subtest was truly multidimensional, since this is a prerequisite for meaningful item classification schemes. The results showed that the subtest was, in fact, multidimensional. The second stage of the study examined the discriminant validity of the classification schemes by determining the number of items within each subskill category that were more highly correlated with the corrected total score of the subskill category of which they were a part than with the total scores for each of the remaining subskill categories. Results indicated that the researcher-developed classification scheme evidenced a 9% increase in the number of items manifesting discriminant validity. The findings suggested that neither classification scheme offered sufficient evidence of discriminant validity to establish its practical utility. (Author/FL)