Osteoclasts express the αvβ3 integrin, an adhesion receptor that has been implicated in bone resorption and that is therefore a potential therapeutic target. To assess the role of this heterodimer in skeletal development in vivo, we engineered mice in which the gene for the β3 integrin subunit was deleted. Bone marrow macrophages derived from these mutants differentiate in vitro into numerous osteoclasts, thus establishing that αvβ3 is not necessary for osteoclast recruitment. Furthermore, the closely related integrin, αvβ5, does not substitute for αvβ3 during cytokine stimulation or authentic osteoclastogenesis. β3 knockout mice, but not their heterozygous littermates, develop histologically and radiographically evident osteosclerosis with age. Despite their increased bone mass, β3-null mice contain 3.5-fold more osteoclasts than do heterozygotes. These mutant osteoclasts are, however, dysfunctional, as evidenced by their reduced ability to resorb whale dentin in vitro and the significant hypocalcemia seen in the knockout mice. The resorptive defect in β3-deficient osteoclasts may reflect absence of matrix-derived intracellular signals, since their cytoskeleton is distinctly abnormal and they fail to spread in vitro, to form actin rings ex vivo, or to form normal ruffled membranes in vivo. Thus, although it is not required for osteoclastogenesis, the integrin αvβ3 is essential for normal osteoclast function.