In the early 1960s I applied 3H-thymidine autoradiography to the study of the cells constituting the neural tube, and found that its wall was composed solely of one kind of single-layered epithelial cell, which perform an elevator movement between the mitotic and DNA-synthetic zones in the wall in accord with the cell cycle. They were identified as multipotent stem cells of the central nervous sytem (CNS) to which I gave the name of matrix cells. (3)H-thymidine autoradiography also revealed the chronology of development of these matrix cells: At first they proliferate only to expand the population (stage I), then switch to differentiate specific neuroblasts in given sequences (stage II), and finally change themselves into ependymoglioblasts, common progenitors of ependymal cells and neuroglia (stage III). Based on these findings, I proposed a monophyletic view of cytogenesis of the central nervous sytem. This matrix cell theory claiming the existence of multipotent stem cells has long been the target of severe criticism and not been accepted among neuro-embryologists for a long time. Recent findings by experimental and clinical neuroscientists on the importance of stem cells have renewed interest in the nature and biology of the multipotent neural stem cells. The present paper describes how the concept of the matrix cell (multipotent neural stem cells in vivo) emerged and what has come out from this view over the last 45 years, and how the basic concept of the matrix cell theory has recently been reconfirmed after a long period of controversy and neglect.