This research investigated the effects of the ability grouping method of between-classroom flexible grouping on intermediate elementary students' mathematical concept improvement and academic self-concepts. This ability grouping method involved relocating students among classrooms within a grade level, based on ability, for mathematics instruction. Pretests, given prior to each mathematics unit, determined students' placements for each unit of study, resulting in the flexible grouping of students into groups throughout the course of the school year. The treatment group, implementing between-classroom flexible grouping for mathematics instruction, consisted of 140 fourth and fifth grade students at two elementary schools. The comparison group, utilizing heterogeneous classrooms, included 142 fourth and fifth grade students at one elementary school. All three schools were situated within one suburban Chicago school district. A quasi-experimental, quantitative research design was employed to collect data. The STAR Math test was administered in the fall (pretest) and in the spring (posttest) of the school year to ascertain mathematical concept improvement. The School Subjects and Mathematics subscales of the Self-Description Questionnaire I were administered to determine students' general academic self-concepts and mathematics self- concepts. At the fourth grade level, the treatment group reported higher mathematical concept improvement, academic self-concepts, and mathematics self-concepts than the comparison group. However, a statistically significant difference was only found for academic self-concepts. At the fifth grade level, the comparison group achieved higher mathematical concept improvement and academic self-concepts. The treatment group attained higher mathematics self-concepts. No statistical significance was determined. When fourth and fifth grades were combined, the treatment group outperformed the comparison group for mathematical concept improvement, academic self-concepts, and mathematics self-concepts. Statistical significance was not attained. In addition, this study attempted to confirm the assertions of Marsh's Big-Fish-Little-Pond Effect. A positive correlation between mathematical ability levels and mathematics self-concepts was noted for all subgroups. A smaller range of mathematics self-concept means across ability levels was evident for the treatment group. The findings of this research suggest between-classroom flexible grouping may benefit intermediate elementary students, but more research is necessary to fully gauge its effects.