It's a very common requirement in development practice that communities contribute towards the cost of projects to provide local public goods and services.  Practitioners argue that requiring communities to contribute ensures that the communities care about the outcomes of the project, making it more likely that they continue to use and maintain the assets created.  More mundanely, community contributions also reduce the cost of project implementation. However, requiring communities to contribute may exclude the poorest communities from participating, and where project decisions are taken collectively, may reinforce existing inequalities within communities, by allowing those who are better able to meet the costs of contribution to exercise greater influence over decision-making processes.  In general, projects that require communities to contribute are not comparable to projects without this requirement, because many other differences bias these comparisons.  In our study, we will randomly assign different contribution requirements to communities who receive an otherwise identical intervention, a project to improve access to safe drinking water.  In this way we will provide the first experimental evidence on whether requiring a community contribution affects the types of communities who decide to participate in the project; which community members meet the cost of the contribution; and how the requirement for a contribution influences the overall impact of the project, in terms of improving access to safe drinking water.