This research evaluated mathematics achievement and attitudinal effects of three instructional methods directed in varying degrees toward accommodating diversity in students' prior achievement. Two randomized field experiments of 16 and 18 weeks' duration, respectively, compared an individualized model, Team Assisted Individualization (TAI); an ability grouped model, Ability Grouped Active Teaching (AGAT); a group-paced model, the Missouri Mathematics Program (MMP); and, in Experiment 2 only, untreated control classes. The first experiment involved pupils in grades 4-6, while the second experiment was conducted with pupils in grades 3-5. Nested analysis of variance on CTBS Computations scores adjusted for pretests indicated that in both experiments, TAI and AGAT exceeded MMP. In Experiment 2, TAI, AGAT, and MMP also exceeded control. No interactions with prior achievement were found. Effects on Liking of Math Class and (in Experiment 1) Self-Concept in Math favored TAI. (Author/MNS)