Twenty years ago, the study by Livingstone and Hubel [(1981) Nature, 291, 554] was viewed as a first step toward understanding how changes in state of vigilance affect sensory processing. Since then, however, very few attempts have been made to progress in this direction. In the present study, 56 cells were recorded in the auditory cortex of adult, undrugged guinea pigs, and the frequency tuning curves were tested during continuous and stable periods of wakefulness and of slow-wave sleep (SWS). Twelve cells were also tested during paradoxical sleep. Over the whole cell population, the response latency, the frequency selectivity and the size of the suprathreshold receptive field were not significantly modified during SWS compared with waking. However, this lack of global effects resulted from the heterogeneity of response changes displayed by cortical cells. During SWS, the receptive field size varied as a function of the changes in evoked responses: it was unchanged for the cells whose evoked responses were not modified (38% of the cells), reduced for the cells whose responses were decreased (48%) and enlarged for the cells whose responses were increased (14%). This profile of changes differs from the prevalent receptive field shrinkage that was observed in the auditory thalamus during SWS [Edeline et al. (2000), J. Neurophysiol., 84, 934]. It also contrasts with the receptive field enlargement that was described under anaesthesia when the EEG spontaneously shifted from a desynchronized to a synchronized pattern [Wörgötter et al. (1998), Nature, 396, 165]. Reasons for these differences are discussed.