During prolonged contractions, recruitment of additional motor units is considered to be a mechanism that compensates for the decrease in force output of the fatigued muscle units. Not much attention has been paid to the effect of prolonged contractions on the motoneurone part of the motor unit. In the following results we demonstrate substitution and rotation among motoneurones which are active over prolonged periods of time. Intramuscular motor unit activity and surface EMG (electromyogram) were recorded from one of the following four muscles: flexor and extensor carpi radialis, tibialis anterior and soleus. The subject was asked to discharge a discernible unit at a comfortable rate with audio and video feedback. The firing rate of this motor unit was perturbed by various phasic synaptic inputs (from the motor cortex or spindle afferents) during the 2-3 h recording period, which was the main purpose of these experiments. Since each motoneurone discharged for a very long time during such experiments, we took the opportunity to examine the effect of prolonged firing on its discharge behaviour by checking for substitution and rotation among low threshold motoneurones. Results are reported from a total of 10 sets of motor units from all four muscles. When a subject fired a motor unit for a long period, frequently an additional motor unit started to discharge after 10-20 min. When the subject was asked to keep activity down to one unit, very frequently, it was unit 1 that dropped and unit 2 continued to fire. After unit 1 silenced, it was not always possible to bring it back immediately by increasing EMG. After a few minutes, one could re-recruit it by increasing voluntary effort. After 5-10 min, unit 1 came back without any conscious effort by the subject. Now if the subject was again asked to retain just one unit, it was unit 2 that dropped. During long experiments, we observed such back and forth rotation of activity in all four muscles. In addition, it was observed that the dropped motor unit also had difficulty responding phasically. Just after it was dropped, it responded phasically to the inputs quite infrequently. As time passed, it responded more frequently until it became tonic again. These observations suggest that the threshold of a tonically firing motoneurone increases with time, which may result from sodium inactivation that has a long time constant. At the same time a newly recruited unit may have a decrease in threshold due to persistent inward currents. These two mechanism may allow the observed substitution and rotation among low threshold motoneurones.