This paper analyzes the effect of incentive-compatible self-chosen goals on academic performance by means of a randomized field experiment. We use two alternative payment mechanisms, a piece-rate and a rank-order tournament, to motivate students depending on their absolute or relative academic performance respectively. Students enrolled in Introductory Microeconomics were classified in two types depending on whether they had a failed background in this course (returning students) or they had not (new students). Controlling for potential confounding factors such as gender, degree, professor and university entrance grade, we find that both payment mechanisms are effective increasing grades of new and returning students.