The way in which preferential votes of a candidate are spread over a large constituency is analysed for the Belgian Senate elections of 1995 and the European elections of 1994 in Belgium, which are both held in large constituencies. A formula that indicates the concentration of preferential votes controls in each sub-unit of the constituency for the number of votes, the number of votes of the candidate's party and the number of preferential votes. When this variabele is combined with a variabele indicating the total number of preferential votes of a candidate, an interesting classification of nine categories of candidates is archieved. The fact that some candidates have a weak concentration of preferential votes in a sub-unit can be explained either by the social distance between candidate and citizen or by the f act that some candidates, due to their media appearences, are welt known all over the constituency and hence that they obtain an equal share of preferential votes in each sub-unit. The place on the party list and the visibility of a candidate in the campaign have an influence upon the category under which a candidate ressorts.