Ferritin was used as a tracer to investigate pathways and mechanisms for transfer across the various layers of the glomerular capillary wall. Kidney tissue, fixed at intervals of 2 minutes to 2 hours following an intravenous injection of ferritin, was examined by electron microscopy. The observations confirmed the existence of three distinct and successive layers in the glomerular capillary wall (the endothelium, the basement membrane, and the visceral epithelium). In addition, they demonstrated a number of new structural features: namely (a) discrete fibrils in the subendothelial spaces; (b) a characteristic, highly elaborate, cytoplasmic organization in the visceral epithelium; and (c) special structures resembling "desmosomes" in the slits between foot processes. In animals sacrificed at short time intervals (2 to 15 minutes) following ferritin administration, ferritin molecules were found at high concentration in the lumen and endothelial fenestrae, at low concentration in the basement membrane, and in very small numbers within the epithelium. Later (1 to 2 hours), the tracer particles were still present in the lumen and within endothelial fenestrae, and, in addition, had accumulated on the luminal side of the basement membrane, especially in the axial regions of the vessels. Larger numbers of ferritin molecules were also found in the epithelium-in invaginations of the cell membrane at the base of the foot processes, and in various membrane-limited bodies (vesicles, multivesicular bodies, vacuoles, and dense bodies) present within the cytoplasm. These observations suggest that the endothelial fenestrae are patent and that the basement membrane is the main filtration barrier. Since the basement membrane has no demonstrable pores, it is probably not a simple sieve but presumably is a gel-like structure with two fine fibrillar components embedded in an amorphous matrix. Both the epithelium and endothelium may be concerned with building and maintaining this structure. Finally, the intracellular accumulation of particles in the epithelium suggests that the latter acts as a monitor that recovers, at least in part, the small amounts of protein which normally leak through the filter.