We have analyzed the diffusion of the Bolsa Família Program, emphasizing the importance of the beneficiary monitoring the Program in their municipality and inducing local governments to commit their efforts to implement the program. This conditional cash transfer program expanded rapidly countrywide and kept its level of targeting the poor comparable to the best international practices. Using a stylized theoretical model, we test whether voters in Brazil monitored the efforts of mayors in implementing the Bolsa Familia Program, rewarding those that were more effective by reelecting them in the 2004 municipal elections. Voters can mitigate political agency problems by comparing the performance of their politicians with those of neighboring municipalities and rewarding or punishing them in their bids for reelection. We employ a quasi-experimental design to analyze the electoral impact of the Bolsa Familia Program during the early stages when several factors that were not directly under the municipalities' control hindered its implementation. We find robust evidence that, in those places where the implementation of the program is lagging behind that of neighboring towns, mayors have a strong incentive to step up their efforts. This result configures an electoral game where incumbent mayors feel obliged to at least match the performance of their peers in order to achieve reelection. This mechanism helps to explain the quick diffusion of the Bolsa Familia Program across the country. In addition, we examined whether electoral incentives influenced mayors to improve their performance in implementing the PBF, assessing whether mayors in their first term (and trying for reelection) outperform their peers in the second term,. Taking advantage of the discontinuities in elections defined by narrow margins of victory, we mitigated the problem of endogeneity by omitted variable and found evidence that mayors in the first term outperformed their second term peers by 12%. Moreover, if the analysis is restricted to municipalities with high electoral competition or with a high proportion of the population in poverty, the mayor's performance in the first term is found to be almost 30% higher. These results were robust to different methods of estimation. In sum, we bring evidence that voters, in possession of relevant information to evaluate the performance of local government, strengthen the control exercised by the federal government on municipalities in implementing the Bolsa Familia Program and create incentives for improved service delivery. Our empirical results are anchored by agency theory with multiple principals: when an agent hired to provide a service is monitored by various parties which share common goals and whose actions are mutually reinforcing, the result is an equilibrium that reduces agent income and/or shirking.