Although it has been recognized for some years that each cytoarchitectonic field of Ammon's horn and the subiculum gives rise to a specific pattern of cortical and subcortical projections, it has not been clear whether these various projections arise from different populations of neurons within each field or whether they arise as collaterals from an essentially homogeneous population of cells. We have examined this problem, and the related issue of the origin of the commissural and ipsilateral associational projections of the dentate gyrus, by injecting retrogradely transported fluorescent dyes into two or more of the relevant projection fields in adult rats and subsequently examining the brains for doubly or triply labeled neurons. It is clear from these experiments that at least two of the known efferent projections of field CA1 (to the septum and to the entorhinal cortex) arise from the same pyramidal neurons and also that the commissural, ipsilateral associational, septal, and subicular projections of the other major field of Ammon's horn--field CA3-- similarly are due to collaterals. Double-labeling experiments also indicate that at least 80% of the cells in the deep hilar region of the dentate gyrus give rise to both an ipsilateral (associational) and a crossed (or commissural) projection to the dentate granule cells. In contrast, the projection neurons in the dorsal part of the subiculum form a quite heterogeneous population; cells that project to both the septum and the entorhinal area are intermingled with others that project to one or the other area but not to both. The cortical and cortico-subcortical connections of the hippocampal formation thus appear to be quite different from those of the neo-cortex, and the existence of such an extensive system of collateral projections clearly has important consequences for studies of the development of the hippocampus and of its response to selective deafferentation.