This study analyzed changes in the performance of a reaching task and its adaptive modification produced by reversible inactivation of three sites within the forelimb representation of the motor cortex (MCx, area 4 gamma) in five cats by microinjections of muscimol. Two sites were located in the lateral MCx, rostral (RL-MCx) and caudal (CL-MCx) to the end of the cruciate sulcus, where intracortical microstimulation (ICMS) produced contraction of the most distal muscles. The third site was located more medially, in the anterior sigmoid gyrus (RM-MCx) where ICMS primarily produced contraction of more proximal muscles. The task required the animals to reach into a horizontal target well, located in front of them at one of three possible heights, to grasp and retrieve a small piece of food. The height of the reach was primarily achieved by elbow flexion. Grasping consisted primarily of digit flexion, and food retrieval consisted of forearm supination and shoulder extension. In some blocks of trials, an obstacle was placed in the path of the limb to assess the animal's ability to adaptively adjust the kinematic characteristics of their response trajectory. In normal animals, contact with the bar on the first trial triggered a corrective response at short latency that allowed the paw to circumvent the bar. On all subsequent trials, the trajectory was adapted to prevent contact with the obstacle, with a safety margin of about 1 cm. Inactivation at all sites produced a slowing of movement, a protracted and extended forelimb posture, and increased variability of initial limb position. In addition, inactivation of RL-MCx immediately produced systematic reaching errors, consisting of hypermetric movements, as well as impaired grasping and food retrieval. The degree of hypermetria was similar for all target heights and was not associated with alterations in trajectory control. During inactivation, animals did not compensate for the hypermetria by reducing paw path elevation, suggesting a defect in kinematic planning or in adaptive control. This was confirmed by finding that trajectory adaptation to avoid bar contact was impaired during RL-MCx inactivation. The short latency corrective response, triggered by contact of the limb with the obstacle was, however, preserved. Inactivation of CL-MCx did not impair aiming, grasping, or adaptation immediately after injection. However, impairments occurred after about 1 h postinjection, and at that time mimicked the effects of RL-MCx inactivation.(ABSTRACT TRUNCATED AT 400 WORDS)