Results of several analyses of the dimensional structure of four forms of the Preliminary American College Test Plus (P-ACT+) are presented. Dimensionality was assessed using factor analysis, multidimensional item response theory (IRT), and the Stout Test for Essential Unidimensionality (STEU). Analyses were conducted with data from the spring 1987 and 1988 equating administrations of the P-ACT+, each with about 1,600 examinees. Factor analyses showed that all tests contained a dominant factor with several additional factors of varying importance. These results were essentially replicated by the multidimensional IRT (MIRT) analyses. The STEU indicated that the tests on each form were not unidimensional. These findings, combined with those from factor analyses and IRT analyses, raise questions concerning the appropriateness of a unidimensional IRT model for score reporting. Consistency of dimensional structure across test forms was investigated by comparing factor patterns, MIRT item statistics, and classical item statistics (p-values, point-biserials, and intercorrelations among scores). Factor patterns and MIRT analyses indicated reasonably consistent patterns in the dimensional structures. Classical analyses also revealed considerable similarity across forms. Use of a unidimensional IRT model will probably provide ability estimates with comparable meaning across test forms. Thirty-four tables and 24 plots present analysis results. There is a five-item list of references. (SLD)