Using a simple experimental design, count data models and routinely collected administrative data from Senegalese health districts, this study examines whether funding community-based organisations can be an effective method of increasing voluntary counselling and testing. Two treatment groups are identified: a newly introduced peermentoring mechanism and traditional social mobilisation. Findings indicate that funded peer mentoring roughly doubles the number of individuals who get tested, who follow pretest counselling and who collect their test results, whereas funded traditional social mobilisation appears to be slightly less effective. Moreover, the study demonstrates that inexpensive administrative data can be used for impact evaluation.