In order to demonstrate the educational benefits of a retraining program on attribution a group of 200 high school girls was randomly selected and their math anxiety measured in order to select the top 40. The selected students were divided into two groups, one of which was put through a retraining program on attribution. In addition to math anxiety, both groups’ attributional styles were identified both before and after the intervention. Both measures were assumed to be of adequate validity and reliability. The data analyses show that the retraining program is effective in changing students’ attributions in both positive and negative situations, as well as in relationship to math anxiety (learning and evaluation). Thus it is suggested such retraining programs be utilized in schools.