Synchronisation of motor unit discharges is commonly seen in hand muscles of normal man but is absent following neurologically complete spinal cord injury and reduced after stroke. These findings support the notion that some corticospinal inputs to motoneurones are shared and contribute to the observed synchrony of discharge. In this study we have examined motor unit discharge in hand muscles below the level of an incomplete spinal cord injury in an attempt to relate strength of synchrony to the integrity of the corticospinal tract. Eight patients with incomplete spinal cord injury (neurological level C3-C7) and eight control subjects took part in the study. The patients had sustained injury 14-191 weeks prior to the recordings and had since regained good motor function in their hands. Two concentric needle electrodes were inserted into the first dorsal interosseus muscle which subjects were instructed to contract weakly so that potentials from individual motor units could be reliably identified on both recordings. Synchrony was detected by constructing cross-correlograms between the discharges of pairs of individual motor units. The amount of synchronous firing was determined from the magnitude of any peak in the cross-correlogram, as the probability above chance (XP) of one motor unit firing with respect to the other and vice versa. The degree of synchrony was lower (P < 0·05) in the patient group (mean XP 0·06) than in the control group (mean XP 0·09). The incidence of significant synchrony was lower in the patient group (41·8 %) than in the control group (92·9 %). The mean (± S.E.M.) frequency of motor unit discharge was slightly lower (P < 0·05) in patients (9·7 ± 0·4 impulses s-1) than controls (10·8 ± 0·5 impulses s-1). The mean width of synchrony peaks was narrower (P < 0·05) in patients (11·4 ± 1·1 ms) than controls (13·2 ± 0·6 ms). We conclude that the weaker synchrony of motor unit discharge in incomplete spinal cord injury may reflect permanent damage to some corticospinal axons.