Extra-pair paternity is an important aspect of reproductive strategies in many species of birds. Given that in most species females control whether fertilization occurs, they are expected to benefit in some way from the extra-pair matings. In this study we use patterns of extra-pair paternity (EPP) in broods of individual reed buntings (Emberiza schoeniclus), both within and between seasons, to test four hypothesized female benefits: (1) assessing potential future partners and seeking (2) genetic diversity (3) good genes, or (4) compatible genes. Reed buntings are socially monogamous, multibrooded passerines with extremely high levels of extra-pair paternity. We studied a population of reed buntings in the Netherlands in 2002 and 2003; 51% of offspring in 74% of nests were extra-pair. We showed that patterns of EPP did not support the first and second hypotheses, since females did not form a pair with previous extra-pair partners, EPP was not evenly distributed among broods and more broods than expected were sired by a single male. Furthermore, there was no relation between a male's within- and extra-pair fertilization success, no consistency in EPP between breeding attempts, no effect of parental relatedness on EPP and several cases of reciprocal paternity. These patterns do not support the good genes hypothesis and are most consistent with the genetic compatibility hypothesis. However, our previous finding that older males are more successful in gaining EPP, suggests some effect of good genes. These hypotheses need not be mutually exclusive, as females may select compatible males above a certain quality threshold (e.g. old males).