The automation of the election process has been experimented in many countries during recent years, to demonstrate that it accelerates the election process and that it offers many advantages; however, such automation also needs to satisfy many security requirements to guarantee a transparent process. In this dissertation, a model for an electronic voting system is proposed. This model focuses on the security risks and the vulnerabilities associated to these processes. As in any election process, electronic voting needs to meet the appropriate standards regarding the basic principles and attributes of a good democratic election. In this study, the principles considered as the basic requirements for electronic voting, are analyzed and included in the proposed model. This dissertation discusses the Brazilian case for being the first country in the world where a 100% of the citizens voted electronically. It also presents other experiences related to Direct Recording Electronic voting in other countries in order to compare and critically analyze the different models. The best features of each model are taken and examined in order to propose secure electronic elections that maintain the selected principles as key requirements. / Dissertation (MIT)--University of Pretoria, 2010. / Computer Science / unrestricted