The short-term dynamics of resetting the circadian 'clock' was assessed by a double-pulse paradigm in vitro. On day 1, single and double 1 h 'pulses' of 1 mM l-glutamate were applied to the rat suprachiamastic nuclei (SCN). On days 2 and 3, single unit activity (SUA) was recorded and time-of-peak SUA was used as a phase marker of the circadian rhythm. The time-of-peak in untreated slices at 'Zeitgeber' time (ZT; hours after lights-on) 6, was used to evaluate effects of glutamate on phase. In accordance with published data, a single glutamate pulse at ZT 14 resulted in a 3 h delay of peak SUA on days 2 and 3. A 2nd pulse, given 3 h after a 1st pulse, resulted in two distinct peaks on day 2: a 1st at ZT 7 and a 2nd at ZT 12, i. e., a 6 h phase delay and hence twice the delay obtained after a single pulse. On day 3, no peak in SUA was observed which indicates that a new steady state was not reached on day 2. The bimodal distribution of SUA on day 2 corroborates other findings which suggest that the SCN comprises two distinct neuronal populations with circadian firing patterns that are normally coupled but, possibly due to different sensitivities to glutamate, can desynchronize. The additive phase-shifting effect of two consecutive glutamate pulses suggests that, at least for one sub-population of SCN neurons, the phase shift is completed within 3 h.