1. Synchronization of concurrently active motor-unit firings was studi
ed in six human muscles performing isometric constant-force contractio
ns at 30% of the maximal level. The myoelectric signal was detected wi
th a quadrifilar needle electrode and was decomposed into its constitu
ent motor-unit action-potential trains with the Precision Decompositio
n technique, whose accuracy has been proven previously. 2. Synchroniza
tion was considered as the tendency of two motor units to fire at fixe
d time intervals with respect to each other more often than would be e
xpected if the motor units fired independently. A rigorous statistical
technique was used to measure the presence of peaks in the cross-inte
rval histogram of pairs of motor-unit action-potential trains. The loc
ation of the center of peak as well as their width and amplitude were
measured. A synch index was developed to measure the percentage of fir
ings that were synchronized. The percentage of concurrently active mot
or-unit pairs that contained synchronized firings was measured. 3. Syn
chronization of motor-unit firings was observed to occur in two modali
ties. The short-term modality was seen as a peak in the cross-interval
histogram centered about zero-time delay (0.5 +/- 2.9 ms, mean +/- SD
) and with an average width of 4.5 +/- 2.5 ms. The long-term modality
was seen as a peak centered at latencies ranging from 8 to 76 ms. On t
he average, the peaks of the long-term synchronization were 36% lower
but had approximately the same width as the peaks for the short-term s
ynchronization. Short-term synchronization was seen in 60% of the moto
r-unit pairs, whereas long-term synchronization was seen in 10% of the
pairs. 4. Short-term synchronization occurred in bursts of consecutiv
e firings, ranging in number from 1 to 10, with 91% of all synchronize
d firings occurring in groups of 1 or 2; and the bursts of discharges
appeared at sporadic times during the contraction. 5. The amount of sy
nchronization in motor-unit pairs was found to be low. In the six musc
les that were tested, an average of 8.0% of all the firings were short
-term synchronized, and an average of 1.0% were long-term synchronized
. The synch index was statistically indistinguishable (P = 0.07-0.89)
among the different muscles and among 9 of the 11 subjects tested. 6.
Sixty percent of concurrently active motor-unit pairs displayed short-
term synchronization, 10% of the pairs displayed long-term synchroniza
tion, and 8% displayed both modalities. Motor-unit synchronization lev
el was significantly lower in the larger muscles [tibialis anterior (5
4% short-term, 13% long-term). and deltoid (45% short-term, 12% long-t
erm)] than in the smaller muscles [first dorsal interosseous (71% shor
t-term, 15% long-term), extensor carpi ulnaris (74% short term, 11% lo
ng-term), and the extensor carpi radialis longus (69% short-term, 9% l
ong-term)], and in 2 of the 11 subjects. 7. The amount of synchronizat
ion was not found to be dependent on the recruitment threshold of the
motor units or on the difference between the recruitment thresholds of
any two motor units that displayed synchronization. 8. Our results ar
gue against the common presynaptic fiber-branches hypothesis and the m
uscle afferent feedback hypothesis that have previously been used to e
xplain the occurrence of synchronization. 9. We suggest that synchroni
zation among motor-unit firings may not have a significant physiologic
al purpose; instead, it may be a by-product of the activity of other p
hysiological mechanisms. 10. We introduce a new hypothesis for explain
ing synchronization. Oscillators within the CNS drive motoneurons to f
ire in synchrony occasionally. This hypothesis requires no direct comm
on physical connection to drive the motoneurons and is capable of expl
aining all our observations.