In a previous study, using co-cultures of embryonic bone rudiments stripped of periosteum, and mononuclear phagocytes of various sources, we found that multinucleated mineral-resorbing osteoclasts developed in vitro from radiosensitive mouse bone marrow mononuclear phagocytes (BMMP). (Burger, E. H., J. W. M. van der Meer, J. S. van de Gevel, C. W. Thesingh, and R. van Furth, 1982, J. Exp. Med. 156:1604-1614. In the present study, this co-culture technique was used to analyze the influence of bone-forming cells on osteoclast formation and bone resorption by BMMP or peritoneal exudate cells (PEC). BMMP or PEC were co-cultured with liver or dead bone, i.e., in the presence or absence of liver boneforming cells. Mineral resorption and osteoclast formation were monitored via 45Ca release from prelabeled live or dead bone followed by histology.Osteoclasts developed from precultured BMMP as indicated by [3H]thymidine labeling, but only in live and not in dead bone. They formed readily from BMMP but only erratically, and after a longer culture period, from PEC. Macrophages from BMMP and PEC invaded live and dead bone rudiments but did not resorb the intact mineralized matrix. In contrast, ground bone powder was resorbed avidly by both cell populations, without formation of osteoclasts.We conclude that live bone-forming cells are required for osteoclast formation from progenitors. Live bone is only resorbed by osteoclasts, and not by macrophages. Osteoclast progenitors are abundant in cultures of BMMP but scarce in PEC, which makes a direct descendance of osteoclasts from mature macrophages unlikely.The nature of the mononuclear precursor cell of the osteoclast and the interactions between bone-forming and bone-resorbing cells are the subject of many recent studies (1, 2). In an earlier paper we presented evidence that multinucleated mineral resorbing osteoclasts will develop in vitro from cultures of proliferating bone marrow mononuclear phagocytes (BMMP) ~ (3). This was demonstrated by co-culturing adult mouse BMMP with noninvaded embryonic mouse long bones that had been stripped of adhering periosteum-perichondrium to remove the endogenous pool of osteoclast precursor cells (4). As the bones were dissected and stripped before cartilage invasion and osteoclast formation had occurred, the stripped bone rudiments essentially contained only chondrocytes and