This research explored the effect of attribution retraining on high school students' academic performance in mathematics. The purpose of the study was to modify students' attributions regarding their achievement in mathematics and to teach them adaptive styles of attributions. Sixty-seven students from grades 10 and 11 conducted the Attribution Questionnaire as a pre-test. Twentytwo students were chosen for the experimental group based on their low score on the questionnaire. The control group which had been selected from the other 45 students was formed by matching their results in mathematics with that of the experimental group. Both the experimental and control groups were equal in number (n=22). The study used a pretest-posttest control group design with matching. Although, random assignment of subjects to experimental and control groups was not performed, the research design is a form of quasi-experimental one. The results demonstrated the positive effect of attribution retraining on students' academic achievement in mathematics and their attribution styles. In adopting adaptive attributions, students in the experimental group were able to view their success as a result of effort and ability rather than luck and task difficulty. It is recommended that the training applied in this research be implemented in developmental guidance programs.