In bone modeling or remodeling, osteoclasts recognize the bone resorption site, however, its mechanism in-vivo remains unknown. We have generated the osteoclast-specific transgenic medaka line showing the specific localization of osteoclasts on the neural and hemal arches, and the opg deficient medaka showing the over-induced osteoclasts on the vertebral body. Here, to study the behavior of pre-osteoclasts, we established c-fms knock-out and fosmid transgenic medaka fish. Our results showed the failure of bone resorption in c-fms-a but not c-fms-b knock-out medaka. To visualize c-fms-a positive cells, the c-fms-a fosmid-GFP transgenic medaka line was generated, showing that as same to neural and hemal arches, the c-fms-a positive cells were localized at the centrum column where no osteoclasts were attached in the wild-type fish. To visualize osteoclast differentiation, we generated the c-fms-a formid-GFP/TRAP promoter-DsRed double transgenic medaka line, indicating change of the fluorescent signal in a cell; from green to yellow and from yellow to red in the long-term time-lapse imaging for 2 days. Moreover, in the c-fms-a knock-out/c-fms-a fosmid-GFP/TRAP promoter-DsRed line, decrease of the number of TRAP positive but not c-fms-a positive cells was shown. Taken together, our data suggested the new regulation systems of osteoclast differentiation in-vivo: the site-specific block of osteoclast differentiation by OPG and pre-existence of pre-osteoclasts at the resorption site.