Intracellular studies reveal that, during slow wave sleep (SWS), the entire cortical network can swing rhythmically between extremely different microstates, ranging from wakefulness-like network activation to functional disconnection in the space of a few hundred milliseconds. This alternation of states also involves the thalamic neurons and is reflected in the EEG by a slow (<1 Hz) oscillation. These rhythmic changes, occurring in the thalamo-cortical circuits during SWS, may have relevant, phasic effects on the transmission and processing of sensory information. However, brain reactivity to sensory stimuli, during SWS, has traditionally been studied by means of sequential averaging, a procedure that necessarily masks any short-term fluctuation of responsiveness. The aim of this study was to provide a dynamic evaluation of brain reactivity to sensory stimuli in naturally sleeping humans. To this aim, single-trial somatosensory evoked potentials (SEPs) were grouped and averaged as a function of the phase of the ongoing sleep slow (<1 Hz) oscillation. This procedure revealed a dynamic profile of responsiveness, which was conditioned by the phase of the spontaneous sleep EEG. Overall, the amplitude of the evoked potential changed sistematically, increasing and approaching wakefulness levels along the negative slope of the EEG oscillation and decaying below SWS average levels along the positive drift. These marked and fast changes of stimulus-correlated electrical activity involved both short (N20) and long latency (P60 and P100) components of SEPs. In addition, the observed short-term response variability appeared to be centrally generated and specifically related to the evolution of the spontaneous oscillatory pattern. The present findings demonstrate that thalamo-cortical processing of sensory information is not stationary in the very short period (approximately 500 ms) during natural SWS.