The purpose of this study was to test the effects of the ARCS model applied math classes on academic achievement of elementary school students with underachievement. The participants were 206 underachievers in total(103 each in experimental and control groups), ranging from 4th to 6th grade, in 24 elementary schools in Seoul(12 schools in experimental group, 12 schools in control group) and 24 teachers instructing the students in their own school. The experiment was intervention of 38 lessons(3 classes per a week) for 14 weeks in after-school classes. Pre-test-post-test control group design and ANCOVA were used. The research procedure was as follows: planning and developing lesson plans for the application of the ARCS model in the math course, verification of content validity, the review committee meeting, teacher training, pre-test, intervention committee meeting, intervention and fidelity evaluation, additional teacher training sessions, and post-test. The findings were as follows: the experimental group (ARCS applied classes) received significantly higher math scores than the control group (traditional classes). Also, the descriptive statistics were examined to identify distinguished features according to grades and groups. As a result, this study could obtain information suggesting what kind of differentiated instructional methods are needed to meet individual students’ needs relating to underachievement. Finally, some recommendations about instructional designs enhancing the math achievement of students at-risk for LD and underachievers were discussed.