This article describes a follow-up analysis of findings from a randomized study that tested the efficacy of a blended version of Enhanced Anchored Instruction (EAI) designed to improve both the computation and problem-solving performances of middle school students with disabilities. The goals of the secondary analysis were to track overall error patterns of students in computing with fractions and to compare the effects of EAI and Business As Usual (BAU) on making these errors. Results showed that students taught with EAI reduced their errors compared to students in BAU classrooms and that reducing the one common error led to improved performance. Error pattern analysis provided clues about how to modify instructional materials for improving computation with fractions.