The effect of carbachol microapplication (4 micrograms/250 nl per 90 s) on the discharge of neurons in the anterodorsal pons of four cats was studied using a newly devised microinjector-microelectrode assembly. Neurons were classified according to the magnitude of their discharge rate increases (or decreases) in physiological desynchronized sleep as desynchronized-on (or desynchronized-off) before injecting carbachol. When carbachol produced a desynchronized sleep-like state only half (15 out of 30) of the desynchronized-on cells were activated (desynchronized-on/desynchronized sleep-like state-on) while the other half were not (desynchronized-on/desynchronized sleep-like state-not on). Compared with the non-activated cells, the desynchronized-on/desynchronized sleep-like state-on cells had three features consistent with playing an active role in desynchronized sleep generation: these cells had a higher mean discharge frequency in desynchronized sleep and higher ratio of discharge frequency in desynchronized sleep compared with wakefulness; they did not fire in phase with electromyogram excitation of neck muscles; and they were concentrated in the short latency desynchronized sleep-like state induction zone described in the companion paper. The three-way correlation between the optimal anatomical site for short latency desynchronized sleep-like state induction, the selective neuronal discharge pattern in desynchronized sleep and the cholinergic activation pattern in the desynchronized sleep-like state suggest that we may have identified a neuronal population that is cholinoceptively activated as part of the physiological mechanism of desynchronized sleep generation.