The spike-wave discharges in epileptic brain can be maintained by one or more sequential restarts, which should have mechanisms similar to those of discharge initiation. To examine this hypothesis an original algorithm for automatically determining restarts was developed. To implement the algorithm the windowed fast Fourier transform with a Hann window was used. The method was to track the frequency consecutive drops on a skeleton constructed from a spectrogram in a selected frequency range (7–12 Hz). The method does not rely on changes in the amplitude of the oscillations, while most of the known methods clearly rely on an increase in power in certain spectral ranges. Discharge restarts were successfully detected and were found to be not different from initial starts in terms of main frequency dynamics. Restarts were diagnosed in about four less often than starts. The existence of restarts confirms that different factors are responsible for maintenance of long spike-wave discharges: some discharges are supported by the special maintenance mechanism, others are continuous chains of short discharges.