Multiple popular math curriculum-based measures have recently been revised in ways that may improve their utility for universal screening. However, the applied use of these tools has yet to be evaluated independently. We conducted a retrospective analysis of the diagnostic accuracy of prior year statewide test results and aimswebPlus math composite scores for predicting proficiency on a statewide math test. Data were collected from students in Grades 3 through 8 in a midsize city district in Texas. We examined the accuracy of screening decisions based on state- or vendor-recommended cut-scores and local cut-scores derived using Receiver Operating Characteristic analyses. When applying the vendor-recommended cut-scores, screening decisions based on aimswebPlus scores resulted in inadequate sensitivity in each grade except 8. Screening decisions based on winter aimswebPlus scores were less sensitive than decisions based on the fall scores. The prior year statewide math test resulted in higher sensitivity than aimswebPlus across all grades with sensitivity reaching adequate levels in Grades 5, 7, and 8. When applying local cut-scores that held sensitivity at desired levels ([greater than or equal to] 0.90), decisions made based on the aimswebPlus and prior year statewide math test generally resulted in adequate specificity. However, these local cut-scores resulted in nearly half of each grade being identified as at-risk. Although aimswebPlus math scores were associated with higher area under the curve (AUC) values when compared to studies of prior versions of math curriculum-based measures (CBMs), the vendor-recommended cut-scores were inadequate for this district. Limitations and future directions for research are discussed.