Abstract Home visiting programmes are increasingly recognized as one of the most effective interventions to improve child health and development in low-income settings. However, the best platforms to deliver such programmes remain unclear. We conducted a randomized controlled trial to test the relative effectiveness of child development agents (CDAs) and community health workers (CHWs) as two possible delivery platforms for early childhood development (ECD) focused home visiting intervention in São Paulo, Brazil. A total of 900 children aged 9–15 months were screened for potential study inclusion between January and March 2015. Children who did not attend crèches at enrolment were included in the trial. Children were randomly assigned to control or to receive biweekly home visits either through a CHW in the areas covered by the Brazilian Family Health Strategy (FHS) or by a newly hired cadre of CDAs in the areas not covered by the FHS. The primary study outcome was children’s development (cognition, motor, language and social emotional skills) assessed after 12 months of intervention with the PRIDI and Caregiver-Reported Early Development Instruments tools. A total of 826 mother-child dyads were enrolled in the trial. In intention-to-treat analysis, neither intervention arm improved study outcomes. In per-protocol (PP) analysis, the CDA programme resulted in a 0.22 standard deviation increase in children’s development (95% confidence interval [0.01–0.43]). The results presented in this study suggest that home visiting programmes have the potential to improve child development among poor urban families in Brazil. However, delivering home visiting interventions through already active CHWs may not be feasible in the Brazilian context and coordination across sectors is essential to effective ECD policies.