Online surveys are becoming more and more popular in the social sciences, for example in electoral research. Plenty of studies have shown that participants who take part in online-surveys differ significantly from participants taking part in face-to-face, or telephone surveys, in terms of their socio-demographic background and political attitudes. Still, since electoral research aims primarily to explain voting behavior, online surveys are deemed useful tools, if the relationship between dependent and independent variables are similar in different survey types, for instance, if vote choice can be described by the same factors, regardless of the survey mode. This paper analyzes these relationships by comparing models of voting behavior which are based on data from different survey modes. Survey data from the German Longitudinal Election Study (GLES) is used since this project simultaneously conducted a face-to-face and telephone survey, as well as online surveys ahead of the 2009 German Federal Election. Hence, these are ideal conditions for comparing different survey modes. Consequently, this paper enables one to evaluate the use of online surveys for empirical electoral research.