Summary This paper investigates whether small-scale water supply systems implemented and operated by water user associations increase access to piped water supply in rural Brazil by more than systems by local governments. Starting from 15% to 16% in the year 2000, access rates in rural areas with water user associations increased to 33.4% in 2010. In areas with local government supply systems, access rates only increased to 24.9%. Based on data from Brazilian census and the national water and sanitation survey, the empirical analysis in this paper shows that the observed difference is effectively due to project-type choice. Additionally it points toward higher accountability as a potential reason for better results of community-based projects. In municipalities where social groups requested a new system before the local government started implementation and therefore public awareness for the project was higher, the increase in access rates is comparable to the increase in municipalities with water user association projects. The same is true if local media is present or political competition in local elections is higher. As the effect of project type on access rates might be confounded by simultaneous drivers of project-type choice and access rates, the quantitative analysis is based on a difference-in-difference estimator in combination with kernel matching to overcome the endogeneity of project type. The treatment effect revealed by this analysis is robust to various specification changes and the robustness checks show no structural differences between treatment and control groups that could bias the results. The calculation of matching weights for the kernel matching is informed by semi-structured interviews with academics and sector experts in Brazil explaining the determinants of project choice. The interviews highlight the political economy behind infrastructure expansion in rural Brazil.