The objective of this article is to examine the effect of a set of variables on the vote of the main candidates (Dilma Rousseff, José Serra and Marina Silva), and on invalid votes in the first round of Brazilian presidential elections in 2010. The research is based on the results of “Estudo Eleitoral Brasileiro” (Brazilian Electoral Study) conducted in 2010. The data was analysed using a multinomial logistic regression model, whose results are also presented through graphics with the predicted probability. Among the selected variables, education, region, religion, party preference, self-positioning in the left-right scale and evaluation of government showed the most significant differences between the candidates. On the other hand, sex, age, and participation in the “Bolsa Família” program have limited impact on the differentiation of candidates.