1. In order to study fusimotor control in reduced preparations, soleus muscle spindle afferents were recorded in premammillary decerebrate cats (n = 15) during crossed extensor reflexes and, after spinalization, during locomotion produced by either clonidine or L-l-3,4-dihydroxyphenylalanine (L-DOPA). The soleus muscle was oscillated sinusoidally (025 mm, 4 Hz) and the afferent mean firing rate and modulation were calculated. An increase in firing rate was assumed to arise from activity in dynamic y-motoneurones (dynamic y-drive) when associated with an increase in modulation to stretching, and in static y-motoneurones (static y-drive) when modulation decreased. 2. At rest in all preparations the firing rate and modulation in primary muscle spindle afferents were generally much higher than after de-efferentation (ventral root section), suggesting a predominant dynamic y-drive. Clonidine decreased and even eliminated this presumed resting y-drive in many afferents, both in the decerebrate (7 of 8) and the spinal (6 of 18) state. This effect on y-drive may account, at least in part, for its suppressive effect on spasticity in humans. 3. When locomotion commenced in clonidine-treated spinal cats, primary afferents generally fired with much higher mean rates (+121 %) and lower sensitivities (-32 %), suggesting a large increase in static y-drive (possibly accompanied by a small decrease in dynamic y-drive). These high rates were usually maintained tonically throughout the step cycle. However, a third of the afferents were silenced during locomotor contractions, and de-efferentation had no significant effect on their firing rates. Thus, for some spindles a-activity can occur without significant y-drive. 4. During locomotion in L-DOPA-treated spinal cats the inferred static y-drive only occurred phasically, coactivated with the EMG, though it could precede the EMG by 100-500 ms. In the flexion phase both the afferent rate and modulation were lower than before locomotion, suggesting a lack of effective y-drive. 5. Crossed extensor reflexes in decerebrate cats also produced a substantial increase in primary afferent firing rate (+187 %) and decrease in sensitivity (-37 %), again suggesting increased static y-drive (n = 18). This y-drive was largely independent of EMG activity and often occurred without a-activity. The mean firing rate of secondary muscle spindle afferents increased significantly during locomotion (with L-DOPA) and crossed extensor reflexes, again indicating increased static y-drive. Clonidine reduced or eliminated the y-drive in seven of eight afferents during crossed extensor reflexes. 6. In conclusion, although there are some common features, such as a predominant static y-drive in all walking preparations, the pattern of static and dynamic y-drive is not closely linked to a-activity under the conditions studied. As well as y-drive without a-activity, we have shown for the first time that a-motoneurones can be activated without significant y-drive to many spindles during behavioural tasks.