Community-driven development is a widely employed development strategy for empowering people to choose their own development priorities, to select their own project leaders, and to monitor the implementation of their projects. It is often assumed that this model results in lower corruption rates. In this paper we take a look at two such projects, the Arid Lands Project in Kenya and the KDP/PNPM project in Indonesia. These projects had strikingly different corruption rates, even though the countries in which they operated had similar corruption perception rates at the beginning of the projects. The goal of this paper is to highlight the specific design elements that may account for the different rates of corruption in these two projects.