Spike trains of long duration were recorded from concurrently active pairs of motor units in the human masseter and tibialis anterior muscles. An innovative analysis technique was used to investigate functional coupling between the motoneurons by plotting the discharge frequency of one motor unit with respect to the firing times of the other (peri-spike frequencygram). Conventional cross-correlograms of discharge times were also constructed for each pair to detect synchronous firing and to compare them with the peri-spike frequencygrams (PSFs). The PSFs were examined with the hypothesis that, if the net common input of the two motoneurons was excitatory, the firing frequency of both units should increase around the time of the synchronous discharge of both units (i.e., time zero in the cross-correlogram). Conversely, if the net common input was inhibitory, the firing frequency of both units should fall around time zero. In 24 out of 37 masseter pairs tested, either one (n = 20) or both (n = 4) units of the pair displayed a statistically significant increase (P < 0.001) in the firing frequency around time zero of the PSF. No significant decrease in the discharge frequency was ever detected in any of the units of the 37 pairs tested. The probability of occurrence of a significant increase in firing rate was found to be significantly higher X2 = 5, P < 0.05) in the pairs with significant synchronous firing (13 out of 15) than in the pairs without (11 out of 22). Moreover, the percentage increase in the discharge frequency was found to be significantly higher in the pairs with significant synchronous activity (1.4%) than in those without (0.74%). In 24 out of 56 tibialis anterior pairs tested, either one (n = 19) or both (n = 5) units of the pairs displayed significant changes (P < 0.001) in discharge frequency around time zero of the PSF. A reduction in the mean firing rate starting about 20 ms before time zero and ending around time zero was most commonly detected in the PSF, the probability of this occurrence being significantly higher (X2 = 6.7, P < 0.01) in the pairs with significant synchronous discharges (18 out of 41) than in the pairs without (1 out of 15). However, the mean percentage fall in the firing frequency during this "inhibitory" phase did not correlate in amplitude with the presence or the absence of significant synchronous activity (-1.4% and -1.1% respectively). In many of the cases, this "inhibitory" period was followed by an "excitatory" period where the firing frequency displayed a significant increase (from about 6 ms to 20 ms past time zero). Such dual changes in the firing rate were only found in the pairs with significant synchronous activity (13 of the 41 pairs). The mean percentage increase in the firing frequency during the "excitatory" phase was significantly higher in the pairs with synchronous firing than in the pair without (1.4% and 0.5% respectively). The specificity of the frequency changes was assessed by performing control analyses where the discharge from one unit was cross-correlated with the discharge from the other unit at a different time (time-shuffled pairs). Synchronisation peaks were never observed in the cross-correlograms of the time-shuffled pairs. Furthermore, none of the 18 time-shuffled pairs in the masseter showed significant frequency changes in the same time period as that seen with the non-shuffled pairs. In the tibialis, only 1 out of 31 time-shuffled pairs showed a significant frequency change in one motor unit, and that was in the opposite direction to the frequency changes found in the non-shuffled pairs. It is suggested that the PSF analysis is an important tool for investigating the functional relationship between individual motoneurons as they fire voluntarily.