Simultaneous recordings were made from fusimotor axons in the central ends of filaments of the masseter nerve, and from masseter and temporalis spindle afferents in the mesencephalic nucleus of the fifth cranial nerve in lightly anaesthetized cats. Fusimotor and alpha-motor units in the masseter nerve were differentiated on the basis of their response to passive ramp and hold stretches applied to the jaw. Spindle afferents were identified as primary or secondary according to their dynamic index after administration of suxamethonium. The activity of a given fusimotor unit during reflex movements of the jaw followed one of two distinct patterns: so-called 'tonic' units showed a general increase in activity during a movement, without detailed relation to lengthening or shortening, while 'modulated' units displayed a striking modulation of their activity with shortening, and were usually silent during subsequent lengthening. Comparison of the simultaneously recorded fusimotor and spindle afferent activity suggests that modulated units may be representative of a population of static fusimotor neurones, and tonic units of a population of dynamic fusimotor neurones. In these lightly anaesthetized animals, both primary and secondary spindle afferents showed increased firing during muscle shortening as well as during lengthening. This increase during shortening is not usually seen in conscious animals and reasons are given for the view that it is due to greater depression of alpha-motor activity than of static fusimotor activity during anaesthesia. The results are discussed in relation to the theories of 'alpha-gamma co-activation' and of 'servo-assistance'; and it is suggested that static fusimotor neurones provide a 'temporal template' of the intended movement, while dynamic fusimotor neurones set the required dynamic sensitivity to deviations from the intended movement pattern.