This study examined the effects of selective inactivation of the cerebellar nuclei in the cat on the control of multijoint trajectories and trajectory adaptation to avoid obstacles. Animals were restrained in a hammock and trained to perform a prehension task in which they reached to grasp a small cube of meat from a narrow food well. To examine trajectory adaptation, reaching was obstructed by placing a horizontal bar in the limb's path. Inactivation was produced by microinjection of the GABA agonist muscimol (0.25-1.0 microg in 1 microL saline). Fastigial nucleus inactivation produced a severe impairment in balance and in head and trunk control but no effect on reaching and grasping. Dentate inactivation slowed movements significantly and produced a significant increase in tip path curvature but did not impair reaching and grasping. Selective inactivation of the anterior and posterior interpositus nuclei did not impair grasping but severely decreased the accuracy of reaching movements and produced different biases in wrist and paw paths. Anterior interpositus inactivation produced movement slowing (wrist speed) and under-reaching to the food well. Wrist and tip paths showed anterior biases and became more curved. Also animals could no longer make anticipatory adjustments in limb kinematics to avoid obstructions but sensory-evoked corrective responses were preserved. Posterior interpositus inactivation produced a significant increase in wrist speed and overreaching. Wrist and tip paths showed a posterior bias and became more curved, although in a different way than during anterior interpositus inactivation. Posterior interpositus inactivation did not impair trajectory adaptation to reach over the obstacle. During inactivation of either interpositus nucleus, all measures of kinematic temporal and spatial variability increased with somewhat greater effects being produced by anterior interpositus inactivation. We discuss our results in relation to the hypothesis that anterior and posterior interpositus have different roles in trajectory control, related possibly to feed-forward use of cutaneous and proprioceptive inputs, respectively. The loss of adaptive reprogramming during anterior interpositus inactivation further suggests a role in motor learning. Comparison with results from our earlier motor cortical study shows that the distinctive impairments produced by inactivation of these two nuclei are similar to those produced by selective inactivation of different zones in the forelimb area of rostral motor cortex. Our findings are consistent with the hypothesis that there are separate functional output channels from the anterior and posterior interpositus nuclei to rostral motor cortex for distinct aspects of trajectory control and, from anterior interpositus alone, for trajectory adaptation.