In this paper we look for the factors determining the relative share of preferential votes of individual candidates. We focus on the district council elections in Antwerp that were held in October 2000. The unique aspect of this elections is that they are without precedent and because of this there are strictly speaking no incumbency-effects. Unlike other research that is based only on administrative data, we use survey data for 612 candidates. In this way we enlarge the explanatory potential of (...) our models. Hence, we were able for instance to check the extent to which membership of voluntary associations expands the individual vote share. We fi d that social capital does matter but surprisingly this effect disappears as soon as ballot position effects are taken into account. (shrink)