This study examined the effects of two training methods (interest-based and traditional) used to improve math word problem performance of middle school students with and without learning disabilities, as measured by the Herr Scale of Mathematical Word Problem Solving Situations. Students were randomly assigned to one of the two treatment groups: the interest-based method or the traditional method. Significant treatment group main effects were found in math word problem performance of students with learning disabilities. Results indicated students with learning disabilities in the interest-based method had higher posttest scores compared to students with learning disabilities in the traditional method. In addition, the posttest scores of students with learning disabilities in the interest-based method were similar to the posttest scores of students without learning disabilities. However, there were no significant differences in the posttest scores of students without disabilities between the treatment groups. Curriculum implications were addressed.