We estimate the impact of incentive strength on achievement under a group-based teacher incentive pay program. The system provides variation in the share of students in a subject-grade that a teacher instructs, which proxies for incentive strength. We find that achievement on incentivized exams, but not nonincentivized exams, improves when incentives strengthen. For the incentivized exams, we find that effects fade out monotonically as a teacher's portion of the group increases to between 20 and 30 percentage and are larger for teachers with low-achieving students. Calculations based off these estimates show modest positive effects of the program overall.