This paper examines various approaches to reducing corruption through a randomized field experiment in Indonesia. The study focuses on over 600 village road projects, where engineers independently estimated the prices and quantities of inputs used, and compared these estimates with official expenditure reports to measure missing expenditures. The results show that increasing the probability of a government audit from 4% to 100% significantly reduced missing expenditures by about 8 percentage points, making audits cost-effective. However, increasing grass-roots participation in monitoring only reduced missing wages, with no effect on missing materials expenditures, which account for three-quarters of total expenditures. This suggests that grass-roots monitoring may be subject to free-rider problems. Overall, the findings indicate that traditional top-down monitoring can play a crucial role in reducing corruption, even in highly corrupt environments.This paper examines various approaches to reducing corruption through a randomized field experiment in Indonesia. The study focuses on over 600 village road projects, where engineers independently estimated the prices and quantities of inputs used, and compared these estimates with official expenditure reports to measure missing expenditures. The results show that increasing the probability of a government audit from 4% to 100% significantly reduced missing expenditures by about 8 percentage points, making audits cost-effective. However, increasing grass-roots participation in monitoring only reduced missing wages, with no effect on missing materials expenditures, which account for three-quarters of total expenditures. This suggests that grass-roots monitoring may be subject to free-rider problems. Overall, the findings indicate that traditional top-down monitoring can play a crucial role in reducing corruption, even in highly corrupt environments.