Elections have become part of the regular political routines in Latin America; yet, the attention given by social scientists to issues of voting behaviour has been wanting. This has become even more true for the specific discussion on the determinants of electoral outcomes. Two basic hypotheses have been advanced in relation to this: one, encompassed in lire notion of ''economic voting'' assumes that the incumbents' economic performance is the best predictor of how electorates will collectively allocate their ballots; in contrast, there is the hypothesis of the election as a ''political referendum'': which sees the return of votes going to the incumbent as a function of the political leadership provided by the ruling party in managing noneconomic issues (as measured by the presidential approval rate). Using original economic and survey research data for 30 presidential elections celebrated in 15 Latin American nations, a quantitative aggregate analysis reveals the effect of the approval rate of the president job as the major driving force underlying the electoral outcomes. Controlled by political variables, inflation, GDP growth rates, and unemployment levels show no significant impact thus endorsing the ''referendum'' hypothesis.