Knowledge of the input-output characteristics of various neuronal types is a necessary first step toward an understanding of cellular events related to waking and sleep. In spite of the oversimplification involved, the dichotomy in terms of type I (long-axoned, output) neurons and type II (short-axoned, local) interneurons is helpful in functionally delineating the neuronal circuits involved in the genesis and epiphenomena of waking and sleep states. The possibility is envisaged that cortical interneurons, which are particularly related to higher neuronal activity and have been found in previous experiments to be more active during sleep than during wakefulness, might be involved in complex integrative processes occurring during certain sleep stages. Electrophysiological criteria for the identification of output cells and interneurons are developed, with emphasis on various possibilities and difficulties involved in recognizing interneurons of the mammalian brain. The high-frequency repetitive activity of interneurons is discussed, together with various possibilities of error to be avoided when interpreting data from bursting cells. Data first show opposite changes in spontaneous and evoked discharges of identified output cells versus putative interneurons recorded from motor and parietal association cortical areas in behaving monkeys and cats during wakefulness (W) compared to sleep with synchronized EEG activity (S): significantly increased rates of spontaneous firing, enhanced antidromic or synaptic responsiveness, associated with shorter periods of inhibition in type I (pyramidal tract, cortico-thalamic and cortico-pontine) cells during W versus significantly decreased frequencies of spontaneous discharge and depression of synaptically elicited reponses of type II cells during W compared to S. These findings are partly explained on the basis of recent iontophoretic studies showing that acetylcholine, viewed as a synaptic transmitter of the arousal system, excites output-type neurons and inhibits high-frequency bursting cells. Comparing W and S to the deepest stage of sleep with desynchronized EEG activity (D) in type I and type II cells revealed that: (a) the increased firing rates of output cells in D, over those in W and S, is substantially due to a tonic excitation during this state, and rapid eye movements (REMs) only contribute to the further increase of discharge frequencies; (b) in contrast, the increased rates of discharge in interneurons during D is entirely ascribable to REM-related firing. On the basis of experiments reporting that increased duration of D has beneficial effects upon retention of information acquired during W, the suggestion is made that increased firing rates of association cortical interneurons during REM epochs of D sleep are an important factor in maintaining the soundness of a memory trace.