When the ascending reticular axonal system is stimulated, the responses of distal structures (e.g., the cerebral cortex) appear to outlast the stimulus; these longlasting effects could reflect the intrinsic nature of the distal structure, or the response could reflect an interaction among the reticular cells which tends to prolong the effects of stimulation. To examine the latter hypothesis, single units with ascending axons (projecting units) were recorded in the cat rostral rhombencephalon in acute experiments conducted under halothane-nitrous oxide anesthesia. Stimulation of areas to or through which axons of reticular neurons projected (midbrain tegmentum and lower tectum, medial thalamus, and basal forebrain) produced a consistent and specific response which was elicited only from these areas: suppression of spontaneous activity which was typically elicited from several areas having ascending axons. One-half of these responses were accompanied by a short latency-single spike synaptic excitation. Stimulating areas more than 1.0 mm from the ascending trajectory never produced this response, whereas the number of responses was directly related to the number of projecting axons identified in any one experiment from a given site. Thus, the predominant effect of stimulating within the ascending axonal trajectory was suppression of spontaneous activity in the projecting units, not an 'en cascade' activation of these units; on the contrary, the only type of excitation encountered was a single, short latency spike. Therefore, any effects of stimulation within the ascending reticular pathway which appear to outlast the stimulus (as previously described in the literature) cannot be ascribed to a reverberating (excitatory) circuit among projecting units. A possible source of the synaptic responses of projecting units is a retrograde activation of collaterals interconnecting the reticular cells. If such interaction exists, it is specifically distributed among cells with ascending axons, as the responses were only observed in a very few units not identified by antidromic excitation; however, other evidence is adduced to support the belief that these few units were projecting units whose axons were beyond the reach of the stimulating electrodes. Futhermore, the axons may be bundled such that units with axons nearest that of a given projecting unit give rise to the most extensive synaptic interactions; the activation of these nearby axons suppresses spontaneous activity, while axons farther away have a greater possibility of being excitatory in nature. Should such a medium for interaction exist, reticular collateral interactions might be seen to exist specifically for the purpose of decreasing the activity of cells destined for similar rostral target structures.