In a response-to-intervention framework, schools typically employ a direct route approach to screening, in which students identified as at risk by a screening process are directly placed into intervention. Direct route approaches require screening decisions to be highly accurate, but few studies examining the predictive validity of reading measures report achieving recommendations for classification accuracy. In this study, two approaches to improving the classification accuracy of predictors of Grade 3 reading performance are compared. Findings indicate that the reliance on single screening measures do not result in high levels of classification accuracy. Classification accuracy improved by 2% when a combination of measures was employed and by 6% when a predicted probability risk index was used. Implications for research and practice are discussed.