Two studies using similar methods in two states investigated the long-term predictive utility of two single-skill early numeracy Curriculum Based Measures (CBMs) and the degree to which they can adequately predict high-stakes test scores. Data were drawn from kindergarten and first-grade students. State standardized assessment data from the "Iowa Test of Basic Skills" (ITBS) and "Missouri Assessment Program" (MAP) were used as criterion measures to evaluate multi-year predictive utility. Logistic regression and Receiver Operating Characteristic (ROC) curves were used to examine the extent to which the measures predicted success or failure on ITBS and MAP. Correlational and logistic regression results indicated the Early Numeracy CBMs effectively predict success or failure for the ITBS and MAP. However, further analysis of the diagnostic utility of the measures suggests that practitioners need to carefully consider specificity, sensitivity, and the degree to which the predictability of common screening measures can be generalized to all state standardized assessments.