The cellular origin and degree of collateralization of the subthalamostriatal, subthalamonigral and subthalamopallidal projections in the squirrel monkey (Saimiri sciureus) were studied using lectin-conjugated horseradish peroxidase (WGA-HRP), Nuclear yellow (NY) and Fast blue (FB) as retrograde tracers. In a first experimental group, WGA-HRP was injected in the left putamen and the right caudate nucleus. Following these injections numerous retrogradely labeled neurons occurred in the dorsolateral two-thirds of the subthalamic nucleus on the putamen-injected side, whereas a smaller number of positive cells were found in the ventromedial third of the same nucleus on the caudate-injected side. In a second experimental group NY was injected in the putamen whereas FB was delivered in the substantia nigra on the same side of the brain. After putaminonigral injections subthalamic cells containing the tracer injected in the putamen (about 75-80% of all retrogradely labeled neurons) occurred in the dorsolateral two-thirds of the nucleus, whereas those containing the tracer injected in the substantia nigra (about 20-25% of all positive subthalamic cells) were confined to the ventromedial third of the structure. Approximately 5-10% of all subthalamic positive neurons were double-labeled following putaminonigral injections. In a third experimental group, NY was injected in the caudate nucleus and FB in the substantia nigra on the same side. After such injections cells retrogradely labeled with NY or FB were present in about equal number and appeared closely intermingled in the ventromedial third of the subthalamic nucleus. Less than 10% of all positive subthalamic neurons were double-labeled following caudatonigral injections. In a fourth experimental group, NY was delivered in the globus pallidus and FB in the substantia nigra on the same side. In these animal cells containing the tracer delivered in the pallidum were about 4 times more numerous than those labeled with the tracer injected in the substantia nigra, and approximately 10-20% of all positive subthalamic neurons were double-labeled following pallidonigral injections. Most of these double-labeled cells occurred in the zone where the two populations of single-labeled cells overlapped. Finally, in a last experimental group, NY was injected in the pedunculopontine nucleus on one side and in the substantia nigra on the other. Following these injections cells containing the tracer delivered in the pedunculopontine nucleus were found to be 5-6 times less numerous than those labeled after substantia nigra injections...