Bone-resorbing osteoclasts are formed through fusion of mononucleated precursors. Their choice of partners during the fusion process remains unclear. We hypothesized that osteoclasts are selective in their choice of fusion partner and that this selectivity is based on heterogeneity among the cells with respect to their maturation stage and their expression and cellular organization of fusion factors. Support for this hypothesis was found from immunofluorescence staining of the osteoclast fusion factors CD47, dendritic cell-specific transmembrane protein (DC-STAMP), and syncytin-1. These stainings revealed heterogeneous localization patterns of all three factors within a given culture of osteoclasts. CD47 was found to be localized primarily in small osteoclasts and preosteoclasts, which were also positive for DC-STAMP but negative for cathepsin K expression. A role of CD47 in the early osteoclast fusion steps was also suggested from experiments with a CD47 blocking antibody, which resulted in an inhibition of the fusion of small osteoclasts. Conversely, blocking of connexin 43 affected the fusion of larger osteoclasts with four or more nuclei. The suggestion that different fusion factors function at different stages of osteoclast fusion supports the idea of heterogeneity in the osteoclast population; our results suggest that osteoclast fusion is indeed based on heterogeneity. Considering the in vivo environment in which osteoclasts develop and fuse, our findings seem very applicable and provide novel, important insight into key issues in bone and fusion research.