A skill-by-treatment interaction (STI) isolates skill deficits and manipulates conditions to match them to student needs. Based on the learning hierarchy, preintervention scores can help predict which intervention will be most successful for an individual student. This study compared the efficacy of a modeling and practice-based decoding intervention for 29 kindergarten and first-grade students. Results suggested that grade was not a significant predictor of which intervention was more effective, but preintervention accuracy in nonsense word fluency was a significant predictor of the more effective intervention, accounted for 68% of the variance, and correctly identified the more effective intervention 88% of the time.