The neuronal mechanism which gives rise to the synchronization of motor-unit discharges has been inferred from an analysis of the interspike intervals of individual motor-unit discharges recorded from the soleus muscle. The motor units were divided into two groups on the basis of characteristic changes in their spike trains. The first group maintained a stationary discharge pattern throughout the process of synchronization with a firing rate of approximately 10 spikes/s. Small unidentified units simultaneously recorded gradually grouped around the individual spikes of the first group motor unit and with this process, high-frequency force oscillation appeared phase-locked with each of grouped discharges. The mean period of force oscillation was almost identical to the mean discharge interval. Therefore, the first group motor unit was considered as a pacemaker of this force-oscillation. The second group motor unit underwent from its initially stationary process to a transitional process characterized by spike dropouts from an otherwise regular spike train. When both groups of motor units were recorded by the same electrode, it was found that the firing rate of the second group motor unit discharges gradually approached that of the first group, and the spikes of the first and the second group motor units occurred near or at the same time. The number of double intervals decreased in a highly predictable fashion with an increase in a firing rate. It was furthermore observed that the spikes of a given motor unit whose discharges interval-to-period ratio is smaller at the beginning of transitional process was entrained to the first group motor-unit discharges with a faster time course than the unit whose discharge interval-to-period ratio is larger. The synchronizing process was described from the relations between shorter discharge interval-to-period ratios and the longer-to-shorter interval ratios obtained at several stages from the beginning of transitional process to the final synchronization. Their relations were best drawn by the second-order regression lines. The faster time course of synchronization was reflected in the larger value of coefficient a in the equation. The results of this and previous study (23) further provided evidence to justify that interaction of motor-unit discharges is responsible for the synchronization. Although the neuronal limiting device of the firing-rate control to approximately 10 spikes/s still remains unsolved, the possibility was considered that a disinhibitory neuronal network first acts to synchronize independently firing motoneurons and leads to the oscillation of the stretch reflex loop. This closed-loop system was considered as a site for the stored motor program and the use of disinhibitory neuronal network was discussed in relation to the Harmon's model of neuromimes.