Long bones from mammals host blood cell formation and contain multiple cell types, including adipocytes. Overabundance of adipocytes in bone marrow is associated with pathological conditions such as age-related osteoporosis or marrow aplasia induced by irradiation or chemotherapy. However, physiological functions of bone marrow adipocytes are poorly documented. Herein, we investigated the consequence of total adipocyte deficiency on bone homeostasis in mice. By generating adipocyte-deficient mice using PPARγ deletion, we found that lipoatrophy leads to dramatic alterations of trabecular and cortical femoral bone. More specifically, cortical bone of these mice is extremely porous and poorly defined due to the excessive presence of active bone-resorbing osteoclasts. Thus, despite the supposed role of PPARγ in osteoclast formation, osteoclastogenesis is highly active in this mouse model. Two independent models of lipoatrophy recapitulated this phenotype, demonstrating that hyperosteoclastogenesis is not intrinsically linked to PPARγ deficiency. We further showed that adiponectin, a cytokine produced by adipocytes and mesenchymal stromal cells, is a potent inhibitor of osteoclastogenesis in vitro and in vivo . Consistently, pharmacological activation of adiponectin receptors by the synthetic agonist AdipoRon inhibits mature osteoclast activity both in mouse and human by blocking podosome formation. This adiponectin-inhibiting action on mature osteoclasts occurs though AMPK activation, thereby demonstrating that even fully differentiated and active osteoclasts are sensitive to adipose-derived signals. Finally, we showed that AdipoRon inhibits bone erosion in vivo in a murine model of inflammatory bone loss. Collectively, these data reveal that adiponectin-producing cells are key regulators of bone homeostasis, and that preserving functional bone marrow adiponectin pathway can improve bone integrity in the context of metabolic and inflammatory disorders.