Primary cilium development along with other components of the centrosome in mammalian cells was analysed ultrastructurally and by immunofluorescent staining with anti-acetylated tubulin antibodies. We categorized two types of primary cilia, nascent cilia that are about 1microm long located inside the cytoplasm, and true primary cilia that are several microm long and protrude from the plasma membrane. The primary cilium is invariably associated with the older centriole of each diplosome, having appendages at the distal end and pericentriolar satellites with cytoplasmic microtubules emanating from them. Only one cilium per cell is formed normally through G(0), S and G(2)phases. However, in some mouse embryo fibroblasts with two mature centrioles, bicilates were seen. Primary cilia were not observed in cultured cells where the mature centriole had no satellites and appendages (Chinese hamster kidney cells, line 237, some clones of l-fibroblasts). In contrast to primary cilia, striated rootlets were found around active and non-active centrioles with the same frequency. In proliferating cultured cells, a primary cilium can be formed several hours after mitosis, in fibroblasts 2-4 h after cell division and in PK cells only during the S-phase. In interphase cells, formation of the primary cilium can be stimulated by the action of metabolic inhibitors and by reversed depolymerization of cytoplasmic microtubules with cold or colcemid treatments. In mouse renal epithelial cells in situ, the centrosome was located near the cell surface and mature centrioles in 80% of the cells had primary cilium protruding into the duct lumen. After cells were explanted and subcultured, the centrosome comes closer to the nucleus and the primary cilium was depolymerized or reduced. Later primary cilia appeared in cells that form islets on the coverslip. However, the centrosome in cultured ciliated cells was always located near the cell nucleus and primary cilium never formed a characteristic distal bulb. A sequence of the developmental stages of the primary cilium is proposed and discussed. We also conclude that functioning primary cilium does not necessarily operate in culture cells, which might explain some of the contradictory data on cell ciliation in vitro reported in the literature.