The origin of ramified microglia has been a longstanding controversial issue, with 4 major schools of thought, which state that they are derived (1) from invasion of mesodermal pial elements, (2) from neuroectodermal matrix cells together with the macroglia, (3) from pericytes, and (4) from invasion of monocytes in early development. This paper is in support of the last-mentioned hypothesis. It is known that ramified microglial cells do not divide under normal circumstances, and since our studies in the corpus callosum have shown that these cells do not appear until the fifth postnatal day, it is reasoned that they must be derived from some preexisting mitotically active cells. The putative precursor is the preponderant amoeboid microglia in the same region. Our experimental studies with the carbon labelling technique have demonstrated for the first time that blood monocytes invade into the early postnatal brain to become amoeboid microglia, which then differentiate into ramified microglia. Just like other tissue macrophages, the monocyte-derived amoeboid microglia exhibit features indicative of phagocytic activities. These include the content of hydrolytic enzymes, uptake of carbon, and a characteristic surface morphology, as seen by scanning electron microscopy. The transformation of amoeboid microglia into ramified microglia, which occurs between the second and third postnatal week, is considered to be a regressive phenomenon, as shown by the diminution of their content of hydrolytic enzymes and the downregulation of membrane antigen. Apart from their primary role as active phagocytes, their involvement in Alzheimer's disease (AD) is evidenced recently by the fact that the cells are specifically marked by antibodies present in the cerebrospinal fluid of AD patients. In conclusion, ramified microglial cells are derived from monocytes, but through an intermediate amoeboid microglia stage as active macrophages in the perinatal period.