We evaluate the impact of joint-liability incentives in the classroom using a randomized field experiment. The instructor design groups of three students in the classroom and provides a premium to their homework's grade only if all three members of the group accomplish some requirements. To isolate the joint liability effect from selfish motivations, we design also an individual incentives treatment. We find that joint-liability incentives impact positively on the grades accomplished in homework and midterm exams both in the experimental courses and in the other courses taken by the students in the semester. Though the positive average effect seems to disappear in the final exams, the overall impact of joint-liability incentives on the academic achievements in the semester is still positive. A drawback of this program is a decrease in the satisfaction with classmates. The significant effectiveness of the peer monitoring developed by the joint liability of group incentives provides novel implications for the design of the grading policies in the classroom and for other social settings where incentives may be based in peer monitoring or joint liability.