Nepotism is an important potential conflict in animal societies. However, clear evidence of nepotism in the rearing of queens in social insects is limited and controversial. In the honey bee, Apis mellifera, multiple mating by queens leads to the presence of many patrilines within each colony. When the colonies reproduce through swarming, workers rear a number of new queens, only a few of which will ultimately head a colony. Workers can potentially increase their inclusive fitness by nepotistically favoring full-sister over half-sister queens during the queen rearing and elimination process. Most studies have focused on interactions between workers and immature queens (eggs and larvae) or adult queens who have exited their queen cells. However, adult queens often remain in their queen cells for up to 1 week after emerging from their pupa. In this situation, workers prevent the queens from emerging, feed them, and protect them from other emerged queens. This stage in queen rearing is therefore one in which nepotism could occur. The current study is the first to investigate the kinship between workers and adult queens who have not emerged from their queen cells. We observed the full suite of behaviors expected during this phase of colony reproduction. Although there was no evidence for nepotism in the worker–queen interactions, there was a nonrandom distribution across patrilines of the workers interacting with the queen cells. In addition, in one colony we found differential treatment of fostered (nonkin)-queen cells. Key words: Apis mellifera, honey bee, nepotism, queen imprisonment, swarming. [Behav Ecol 16:403–409 (2005)] I nsect societies are typically nonclonal, and this leads to a wide range of potential reproductive conflicts, including conflicts over sex ratio (Trivers and Hare, 1976), male production (Ratnieks, 1988), caste fate (Wenseleers and Ratnieks, 2004), and queen rearing (Visscher, 1993). Nepo-tism should play an important role in the resolution of these conflicts. However, clear evidence of nepotism in social insects is limited and controversial (see Hannonen and Sundström, 2003, for an example in ants). Multiple mating by honey bee (Apis mellifera) queens leads to the presence of many patrilines (paternal subfamilies within the single matriline) within each colony (Estoup et al., 1994; Palmer and Oldroyd, 2000). This causes potential conflict over queen rearing because it creates relatedness asymmetries between the workers who rear the queens and the young queens themselves, with workers being either full sisters (r ¼ .75) or half sisters (r ¼ .25) to these queens (Visscher, 1986). When colonies reproduce through swarming, workers rear approximately 10–20 new queens, but only a few of these (one to four) will ultimately head a colony. Workers can potentially increase their inclusive fitness if they nepotistically favor full-sister queens or disfavor half-sister queens during the queen rearing and elimination process (Visscher, 1998). However, if workers' recognition ability of full-sister queens against half-sister queens is error prone or if nepotism is costly to the colony as a whole because it reduces queen quality or number, then nepotism would not be selected or would be selected for only the weakly (Ratnieks and Reeve, 1991; Tarpy et al., 2004). When the primary swarm leaves with the old (mother) queen, the new (sister) queens are immature. When these queens mature into adults, they compete to head a new colony either by leaving with a secondary swarm or by becoming the queen in the established nest site. In both cases, the interests of individual queens may be different from those of the workers (Visscher, 1993). The process of queen elimination has been described extensively (Bruinsma et al., 1981; Butler, 1623; Fletcher, 1978; Grooters, 1987), and theoretical work shows that it could be an important stage for queen-queen, worker-queen, and worker-worker conflict (Visscher, 1993). However, the precise role and importance of worker behavior in the outcome of the process have only recently been studied in detail (reviewed in Tarpy et al., 2004). The elimination process is characterized by numerous queen-queen interactions, including queen fights in the form of ''duels'' between adult queens, ''assassinations'' in which a pupal queen in her cell is killed by an adult queen free in the colony, vibratory signals made by adult queens (piping), and queen-worker interactions (vibratory signals, aggressive be-havior, and feeding). These interactions suggest that workers could play an important role in the queen selection process, motivated by either nepotism or ''quality control'' uncon-nected with nepotism (Tarpy et al., 2004). Tarpy and Fletcher (1998) found that queens who were sisters of the workers had an advantage in winning duels over unrelated queens. However, Gilley (2003) found that in colonies with naturally mated queens, aggressive behavior by workers was not more directed toward half-sister queens. Queen quality has little influence on worker-queen interactions and survival (Gilley et al., 2003; Schneider and DeGrandi-Hoffman, 2003; Tarpy et al., 2000). Despite these many studies, our understanding of the queen elimination process remains incomplete. All studies of the influence of workers on the selection of new queens have focused on interactions between workers and immature queens (Châline et al., 2003; Noonan, 1986; Page et al., 1989; Schneider and DeGrandi-Hoffman, 2002; Visscher, 1998) or between workers and adult queens who have exited their special queen cells (Gilley, 2001, 2003; Tarpy and Fletcher, 1998). However, adult queens often remain in their queen cells for up to 1 week (Bruinsma et al., 1981; Fletcher, 1978; Grooters, 1987) before exiting into the colony. During this time, workers cluster on each cell containing an