It is widely believed that the adoption of quality public sector accounting practices can assist in reducing corruption. In theory, accounting reform, especially the shift from cash-to accrual-based methods, leads to the production of much improved financial information, which ultimately can be used by citizens to hold government more accountable and limit corruption. Empirical evidence from cross-country analyses appears to support the theoretical predictions. We investigate the impact of accounting practices on corruption among districts in Indonesia. We use external financial audits to measure the adoption of reforms and the number of corruption case court convictions as our proxy for corruption. We estimate Poisson regression models using instrumental variable techniques to identify the causal effects of adopting accrual-based accounting procedures on corruption. We show that the employment of improved accounting methods is strongly associated with declining corruption. However, after accommodating the endogeneity of accounting practices, we determine that reform adoption has no effect on corruption.