Fifth- and sixth-grade students in one school were instructed using the Mathematics Target System, an individualized approach, while those at a comparable school were given more traditional instruction. Students in both the experimental and control groups were pretested using the Pupil Opinionnaire, Attitude Toward Mathematics scale, and the Stanford Achievement Test. After a year's instruction, students were again tested using parallel forms of these instruments. Differences between groups at each test administration and differences between pre- and post-test means were submitted to analyses using the Chi-square statistic. Although there were some differences between the groups at the outset of the study, the individualized approach was shown to be quite effective with fifth-grade students. The authors observe that independent influences on the control group resulting from increased systemwide emphasis on computational skills and curriculum supervision tended to enhance the learning of students in the control group. (SD)