Models of Response to Intervention (RTI) include parameters of assessment and instruction. This study focuses on assessment with the purpose of developing a screening battery that validly and efficiently identifies first-grade children at risk for reading problems. In an RTI model, these children would be candidates for early intervention. We examined accuracy, fluency, growth, and teacher rating measures as predictors of child status (at risk, not at risk) at the end of the school year based on an unselected sample of 243 children. The prediction model that best fit our selection criteria included 2 word-fluency measures and a teacher rating of reading problems. Word-fluency growth was an equally plausible choice statistically, but, because the measure would require an additional data point, it was not the most efficient choice. The receiver-operator characteristic curve analysis yielded an area-under-the-curve index of 0.96, which indicates the selected 3-variable model is highly accurate. (Contains 1 figure, 4 tables, and 1 note.)