In list PR systems, the criteria applied in the candidate selection process can be analysed by contrasting the realistic and marginal candidates with the non-realistic candidates on the list. Realistic candidates stand a large chance, for marginal candidates this chance is more dependent on their personal result or the result of the list. This approach was applied to the Belgian Chamber elections in the period 1999-2010. A multilevel multinomial regression analysis shows that incumbents, mayors and party employees stand a significantly larger chance of being selected as a realistic or marginal candidate. No evidence was found of a specific gender bias. Once the above-mentioned variables are taken into account, previous campaign experience has a negative effect on the odds of being selected. For the marginal positions, parties particularly seek local office holders from larger municipalities.