To what extent do Voting Advice Applications (VAA) have an influence on voting behaviour and to what extent should providers be hold accountable for such tools? This paper puts forward some empirical evidence from the Swiss VAA smartvote. The enormous popularity of smartvote in the last national elections in 2007 and the feedback of users and candidates let us come to the conclusion that smartvote is more than a toy and likely to have an influence on the voting decisions. Since Swiss citizens not only vote for parties but also for candidates, and the voting recommendation of smartvote is based on the political positions of the candidates, smartvote turns out to be particularly helpful. Political scientists must not keep their hands off such tools. Scientific research is needed to understand their functioning and possibilities to manipulate elections. On the bases of a legal study we come to the conclusion, that a science driven way of setting up such tools is essential for their legitimacy. However, we do not believe that there is a single best way of setting up such a tool and rather support a market like solution with different competing tools, provided they meet minimal standards like transparency and equal access for all parties and candidates. Once the process of selecting candidates and parties are directly linked to the act of voting, all these questions will become even more salient.