Osteoclast formation in the microphthalmic osteopetrotic (mi) mouse was studied from very early embryonic to newborn stages. Embryonic and fetal milmi osteoclasts, generated during the period before bone marrow is formed in the long bones, were predominantly mononuclear and lacked ruffled borders. These cells did, however, show many osteoclastic morphologic and functional properties, such as an abundance of mitochondria, positive succinic dehydrogenase and acid phosphatase reactions, and close contact with and resorption of the calcified cartilage matrix (though diminished). These osteoclastic mononuclear cells appeared in vivo as well as in organ cultures of fetal metatarsal bones with their intact periostea. They also were observed in cocultures of periosteum-free fetal metatarsal bones, with several extraneous sources of osteoclast precursors: yolk sacs and abdominal regions of 9- and 11-day-old embryos, fetal livers, and precultured mononuclear phagocytes isolated from the fetal liver. In contrast, +/+ osteoclasts were always multinuclear, functioned normally in resorbing the calcified cartilage matrix, and had ruffled borders in vivo as well as when derived from the above-mentioned sources. Fetal liver-derived milmi macrophages also failed to form multinuclear foreign body giant cells as opposed to +/+ macrophages in granulomas on implanted pieces of Melinex. The fusion failure of cells derived from embryonic and fetal extramedullary milmi monocyte/macrophage sources contrasted with the occurrence of multinuclear osteoclasts and foreign body giant cells derived from precursors from the bone marrow in young milmi mice. We conclude that the fusion defect of milmi osteoclast precursor cells is already present in their ancestry in blood cell-forming organs of very young embryos and that these cells differentiate into mononuclear osteoclasts that function inefficiently in prenatal bone. We presume that in fully developed bone marrow, local factors are favorable for abolishing the fusion defect.