Genetic dissection of human recessive osteopetroses (ARO) has identified specific subsets due to a defect in molecules linked to the effector function of mature osteoclasts. While an impairment in osteoclast differentiation in mouse leads to osteopetrosis, the four genes identified so far in classical human ARO (TCIRG1, CLCN7, OSTM1 and PLEKHM1) are all involved in the resorption and/or intracellular traffic of the minerals solubilised from bone matrix. The recent finding that the RANKL gene is mutated in a subset of ARO patients whose biopsies did not show any osteoclast shows that a differentiation defect can be responsible for human ARO and paves the way to a potential rational therapy of this rare disease by soluble RANKL administration.