This study aimed to investigate the influence of an attribution training program for learners who attribute their sport performance to dysfunctional attributions. Participants were 35 college beginner tennis players who were oriented to attribute their performance in a tennis skill task to controllable, unstable factors; uncontrollable, stable factors; or no specific factors. Participants received fictitious failure feedback over 10 trial blocks administered during four sessions. Dependent variables included attributions, expectations, emotions, persistence, and performance. MANOVA analyses revealed that it is possible to modify attributions in regard to a tennis performance task. More importantly, the new attributions were consistent up to 3 weeks postintervention and were generalized to a different tennis task. In addition, participants who changed their attributions to more functional ones had higher expectations for future success and experienced positive emotions.