Using a simple experimental design, Poisson regression techniques and routinely-collected administrative data from health districts in Senegal, we examine whether funding community based organizations can be an eective manner of increasing voluntary testing and counselling and modifying the subsequent behavior of individuals who test HIV-positive. We distinguish between two treatment groups: in a rst set of randomly-chosen health districts, community organizations received funding and carried out HIV/AIDS sensitization using traditional social mobilization techniques, whereas in a second randomly-chosen treatment group, they did so by using a new peer-mentoring mechanism; the remaining health districts were assigned to the control group and received no funding. Our results indicate that: funded peer mentoring doubles the number of individuals who get tested, who follow pre-test counselling and who pick up their test results, whereas funded traditional social mobilization appears to be ineective; (ii) both traditional social mobilization and peer mentoring increase the number of partners of HIV-positive individuals who get tested, and (iii) they signicantly increase the number of HIV-positive individuals who follow post-test counselling. Thus, instead of conning attention to a single manner of deploying sensitization campaigns, a hybrid approach, in which peer-mentoring and traditional methods are combined, is advisable. Moreover, inexpensive administrative data can be used for rigorous impact evaluations.