This paper assesses empirically the electoral impact of an immensely popular Vote Advice Application and TV show (‘Do the Vote Test’) during the 2004 Belgian election campaign. Vote Advice Applications are becoming more popular in several Western countries and ever more voters get a voting advice during an election campaign. Drawing on a large panel of Internet users, the study systematically compares users and non-users of this Vote Advice Application, testing whether getting a personal vote advice made any difference. We find that ‘Do the Vote Test’ indeed has affected Belgian voters’ final decision but at the same time these effects were modest. Some parties gained some votes due to the ‘Do the Vote Test’, and others lost some votes, but probably the application did not strongly affect the overall election outcomes. Finally, we show that people's subjective perceptions of the impact of ‘Do the Vote Test’ on their actual electoral behaviour are often contradictory.