This paper examines how self-evaluation biases may influence satisfaction with performance. A review of theoretical positions suggests there are two views, both of which are supported by studies involving laboratory tasks. The first view predicts affective costs, and the second affective benefits of positive self-evaluation bias. We test the opposing predictions with the data of a study in which fifth-graders' self-evaluation, performance, and satisfaction with performance were measured before and after a four-week arithmetic training program. Our findings support the benefit view, that is, students with a negative self-evaluation bias rated their satisfaction with performance significantly less than those with positive bias, even though both performed comparably.