The arborization pattern and postsynaptic targets of corticofugal axons in basal forebrain areas have been studied by the combination of anatomical tract-tracing and pre- and postembedding immunocytochemistry. The anterograde neuronal tracer Phaseolus vulgaris leucoagglutinin was iontophoretically delivered into different neocortical (frontal, parietal, occipital), allocortical (piriform) and mesocortical (insular, prefrontal) areas in rats. To identify the transmitter phenotype in pre- or postsynaptic elements, the tracer staining was combined with immunolabeling for either glutamate or GABA, or with immunolabeling for choline acetyltransferase or parvalbumin. Tracer injections into medial and ventral prefrontal areas gave rise to dense terminal arborizations in extended basal forebrain areas, particularly in the horizontal limb of the diagonal band and the region ventral to it. Terminals were also found to a lesser extent in the ventral part of the substantia innominata and in ventral pallidal areas adjoining ventral striatal territories. Similarly, labeled fibers from the piriform and insular cortices were found to reach lateral and ventral parts of the substantia innominata, where terminal varicosities were evident. In contrast, descending fibers from neocortical areas were smooth, devoid of terminal varicosities, and restricted to the myelinated fascicles of the internal capsule en route to more caudal targets. Ultrastructural studies obtained indicated that corticofugal axon terminals in the basal forebrain areas form synaptic contact primarily with dendritic spines or small dendritic branches (89%); the remaining axon terminals established synapses with dendritic shafts. All tracer labeled axon terminals were immunonegative for GABA, and in the cases investigated, were found to contain glutamate immunoreactivity. In material stained for the anterograde tracer and choline acetyltransferase, a total of 63 Phaseolus vulgaris leucoagglutinin varicosities closely associated with cholinergic profiles were selected for electron microscopic analysis. From this material, 37 varicosities were identified as establishing asymmetric synaptic contacts with neurons that were immunonegative for choline acetyltransferase, including spines and small dendrites (87%) or dendritic shafts (13%). Unequivocal evidence for synaptic interactions between tracer labeled terminals and cholinergic profiles could not be obtained in the remaining cases. From material stained for the anterograde tracer and parvalbumin, 40% of the labeled terminals investigated were found to establish synapses with parvalbumin-positive elements; these contacts were on dendritic shafts and were of the asymmetrical type. The present data suggest that corticofugal axons innervate forebrain neurons that are primarily inhibitory and non-cholinergic; local forebrain axonal arborizations of these cells may represent a mechanism by which prefrontal cortical areas control basal forebrain cholinergic neurons outside the traditional boundaries of pallidal areas.