We study the effect of self-set goals on performance using a randomized field experiment involving first year university students. Students have private meetings with a tutor during the year. We instructed a random group of tutors to induce students to set themselves a course specific goal during the private meeting. A random subset of tutors was further instructed to challenge students to increase this self-set goal. We find that students whose tutor was instructed to induce goalsetting do not perform better compared to the control group but are less likely to drop out of the course by 3%. Students whose tutor was instructed to challenge their students do not differ from the control group in terms of both grade and drop out rate. For those students who actually set a goal we find an increase of 0.2 points (15% of a st.dev) on the grade, and a decrease of 5% points on the drop out rate. Those students who actually set a goal and are challenged score 0.5 points lower, and have a 9% points higher drop out rate compared to students who set a goal but were not challenged.