This paper analyzes the reverse coattail effect on Brazilian elections, a term originally coined by Ames (1994). More specifically, it deals with the ability of local party organizations to transfer votes to upper levels party candidates by concentrating on the causal electoral effect of electing a mayor over subsequent statewide proportional elections. To identify the effect of electing a mayor, it employs a regression discontinuity design (RDD) focusing on observations in which the electoral difference between the elected mayor and the runner-up is very tight. The use of a large dataset, covering elections between 1996 and 2010, allows exploring parties’ heterogeneity inThis paper analyzes the reverse coattail effect on Brazilian elections, a term originally coined by Ames (1994). More specifically, it deals with the ability of local party organizations to transfer votes to upper levels party candidates by concentrating on the causal electoral effect of electing a mayor over subsequent statewide proportional elections. To identify the effect of electing a mayor, it employs a regression discontinuity design (RDD) focusing on observations in which the electoral difference between the elected mayor and the runner-up is very tight. The use of a large dataset, covering elections between 1996 and 2010, allows exploring parties’ heterogeneity in both cross-section and temporal analyses. Main results show that the positive effect of electing a mayor on party performance in subsequent proportional election holds for the entire period. Another finding is that Brazilian parties show different capabilities in getting votes from their mayors, pointing that intra-party linkages may vary among parties. Finally, the impact of electing a mayor vary along the years.