The classical study revealed the melanocyte density in the epidermis of many skin sites of humans. However, these data were obtained by counting melanocytes using dihydroxyphenylalanine (dopa)-treated epidermal sheets. Since rete ridges are developed well in human epidermis, there is a concern about the accuracy of these data. The accurate counting of the melanocyte density in the epidermis including the rete ridges should be performed using histological sections after the dopa treatment. Moreover, it is not known well how many melanoblasts are present in Asian epidermis. The aim of this study was to count the accurate number of melanocytes and melanoblasts. Normal skin sites of 9- to 77-year-old patients were fixed with buffered formalin and processed to the dopa and combined dopa-premelanin reactions. The numbers of cells positive to the dopa reaction (melanocytes) and to the combined dopa-premelanin reaction (melanoblasts and melanocytes) were scored. In the skin of arms, legs, back, and belly, similar density (approximately 110-120cells/0.1mm2 ) of melanocytes was observed, whereas in the skin of scalp, melanocyte density was much lower (approximately 70cells/0.1mm2 ). By contrast, the melanoblast density did not differ between skin sites (approximately 100cells/0.1mm2 ). These results suggest that the melanocyte density does not differ between skin sites except the scalp skin, and a certain number of melanoblasts are present in each skin site of Asian. Melanoblasts seem to be required for producing new melanocytes required to maintain epidermal homeostasis.