The psychometric integrity of selected items from the Mathematics Association of America (MAA) placement tests for college students was investigated. Two alternative and parallel versions of the test were developed (Form A and Form B) for this study. Data for 539 students seeking admission into an undergraduate mathematics curriculum at a private university in the southern United States who completed Form A of the MAA Placement Test were used, as well as data for two sets of the students (N=112; N=196) who completed a retesting using a parallel set of MAA items (Form B). Reliability coefficients for Forms A and B, item difficulty and discrimination, and test-retest reliability coefficients were computed. Results suggest that the measures performed reasonably well, although use of more items may be warranted to yield more psychometrically defensible placement scores. The reliability coefficient for results on Form A suggests reasonable psychometric integrity, but that extreme caution is needed for subscale scores. Subscales on the second version (Form B) also must be interpreted with caution. Two tables summarize item statistics. (SLD)