The original motivation for this study was the observation in previous human experiments that the motor neuron firing frequency (recorded from the motor units in the EMG) was lower at de-recruitment than at recruitment with slow linearly varying voluntary contractions. Is the lower firing rates at de-recruitment correlated with a change in the post-spike afterhyperpolarization (AHP) after preceding spike trains? This question was investigated by intracellular recordings from cat motor neurons in unanaesthetized and anaesthetized preparations. The firing frequency at recruitment and de-recruitment was compared during slow triangular current injections mimicking the slow linearly varying voluntary contractions in humans. There was a lower frequency at de-recruitment in almost all motor neurons (83 of 86 motor neurons; mean =3.35 Hz). Thus intrinsic mechanisms play an important role for the lower frequencies at de-recruitment. This was independent of whether the current injection had activated persistent inward current (plateau potentials, secondary range firing). It was found that a preceding spike train could prolong the AHP duration following a subsequent spike. The lower rate at derecruitment matches the prolongation of the AHP. However, a quantitative comparison between the lowest firing frequency and AHP duration for individual motor neurons reveal that the predicted lowest firing frequency does not match the absolute AHP duration; the lowest frequency is lower than predicted from AHP duration in fast motoneurons, and higher than expected in slow motoneurons. It is suggested that these deviations are explained by the presence of synaptic noise as well as recruitment of persistant inward currents below firing threshold. Thus synaptic noise may allow spike discharge even after the end of the AHP in "fast" motor neurons, while synaptic noise and persistent inward currents below spike threshold tend to give higher minimum firing frequencies in "slow" motor neurons than predicted from AHP duration.