Most development practitioners would list engaging communities in the provision of public services among best practices for improving access. However, whether community participation enhances provision and use of public services relative to a non-participatory approach is largely unknown because few studies compare impacts when the same public service intervention is implemented with and without community participation. This field experiment compares three approaches to providing safe water in rural Bangladesh. Delegating decisions to the community increases use of safe water by about 80% relative to a top-down provider making the same decisions but only when the approach to delegating decisions limits elite influence.