Zoonotic visceral leishmaniasis in Brazil is caused by Leishmania infantum parasites and is transmitted by sand flies of the Phlebotominae family. Dogs are the main urban reservoirs and represent the major source of contagion for the vectors. Studies have shown that most infected dogs are polymerase chain reaction-positive months before seroconversion. Herein, we describe a cohort study designed to identify the incidence of and risk factors for L. infantum infection as detected by polymerase chain reaction-restriction fragment length polymorphism. To determine the risk factors for infection, we conducted a baseline canine survey (n=1443) from which dogs were selected for the cohort study (n=282) involving three evaluations over the course of a 26-month follow-up period. Serology, molecular tests, and a structured questionnaire were used. The risk factors for infection were identified by means of the Cox regression model. The overall infection incidence was 5.8 per 100 dog-months (95% confidence interval 5.1-6.5). Increased risk of infection was associated with the presence of previous cases of canine visceral leishmaniasis in the domiciles (hazard ratio [HR] 1.4; 95% confidence interval [CI] 1.1-1.8) and unplastered house walls (HR 3.6; 95% CI 1.6-8.1). These risk factors suggest that insecticide spraying in cracks and crevices in unplastered walls can reduce biting rates within and around homes. Furthermore, our results demonstrate that the Visceral Leishmaniasis Control and Surveillance Program should adopt environmental management measures in homes with previous cases of canine visceral leishmaniasis, because these homes are more likely to maintain the transmission cycle.