This article analyzes the effects of majoritarian formulas on the effective number of candidates in the elections for president (absolute majority) and mayor (relative majority) in Chile. Although, in theory, greater fragmentation with the absolute majority system with run-off voting would be expected, the authors demonstrate that, in the case of Chile, precisely the opposite is observed. Fragmentation is greater in the relative majority system. This is explained by the lower entry and information costs that exist in mayoral elections with respect to presidential ones. We support this hypothesis with a time series statistical model (cross-sectional time series regression model). We use electoral data added to the communal level for the 2004 and 2008 municipal elections, and for the 1999 and 2005 presidential elections.