Although the Wechsler Intelligence Scale for Children-Revised (WISC-R) is being rapidly replaced by the third edition of the WISC, questions concerning the construct validity of the WISC-R have not yet been resolved, including the number of factors it measures and whether the same constructs fit across all age levels. This study sought to determine whether the WISC-R measures the same constructs across age levels, what constructs it does measure, and how many constructs provide the best fitting model. Multi-sample, hierarchical confirmatory factor analyses using the LISREL computer program (version 7.2) were performed on the WISC-R standardization data. This sample consisted of 2,200 subjects, 200 in each of 11 age groups from 6.5 to 16.5 years. The covariance matrices for the 11 age levels were statistically indistinguishable (p>.05). The test did measure the same constructs across ages. The three-factor model provided a statistically better fit than the two-factor model, and a more parsimonious fit than the four-factor model. In addition, the three-factor model produced a consistently good fit as tested by chi-square holding both measurement and error matrices invariant across all 11 age groups. (Contains 3 tables, 4 figures, and 11 references.) (Author/SLD)