In many countries, college-bound high school seniors must pass a test or series of tests. In Israel, this requirement is known as the ?Bagrut?, or matriculation certificate, obtained by passing a series of subject tests. In spite of the Bagrut?s value, Israeli society is marked by vast differences in Bagrut rates by region and socioeconomic status. We attempted to increase the likelihood of Bagrut certification among low-achieving students by offering substantial cash incentives to high school seniors in an experimental demonstration program. As a theoretical matter, such incentives may be helpful if low-achieving students reduce investment in schooling because of high discount rates, part-time work, or face peer pressure not to study. The experiment studied here used a school-based randomization design offering awards to all students in treated schools who passed their exams. Randomization was imperfect because of the clustered design. We discuss alternative strategies for dealing with clustering in research of this type. On balance, the estimates point to a substantial and statistically significant treatment effect for students close to the margin for certification. We also look at a number of mediating outcomes in an effort to determine how students responded to incentives. These results show students took more tests and were more likely to accumulate the number of credit units required for Bagrut success.