This study examined changes in the performance of a single-joint, elbow task produced by reversible inactivation of local regions within the proximal forelimb representation in area 4 gamma of motor cortex (MCx) and the red nucleus (RN) of the cat. Inactivation was carried out by microinjecting lidocaine, gamma-aminobutyric acid, or muscimol into sites where microstimulation evoked contraction of elbow muscles. Reaction time, amplitude, and speed (velocity or dF/dt) of position and force responses elicited during inactivation were compared to control values obtained immediately prior to inactivation. In addition, we assessed qualitatively the effects of inactivation on reaching, placing reactions, and proprioceptive responses to imposed limb displacement. In the single-joint task, injections in MCx did not increase reaction time (simple or choice) and produced modest and inconsistent reductions in response amplitude (mean -8%) and speed (mean -19%). In contrast, injections of the same amounts of inactivating agents in the forelimb representation of RN consistently increased reaction time (34.4%), and increased the reaction time coefficient of variability (32%). There were small reductions in response amplitude (-4%) and speed (-10%) which were less than those produced by MCx inactivation. During reaching, however, these same injections in MCx and RN produced a substantial loss of accuracy. For MCx, this was due, in part, to systematic hypometria: for RN, inaccuracy resulted from increased variability in paw paths. Placing reactions and corrective responses to imposed limb displacements were also depressed by the cortical and rubral injections. Our results suggest that the forelimb representation in RN plays a role in the initiation of the single-joint, elbow tracking response examined here. The RN may mediate cerebellar regulation of response timing, a function that is likely to be important for interjoint coordination. Although neurons in the forelimb representations of MCx may contribute to force generation in single-joint movements, their contribution to multijoint control appears to be more important and is examined in the subsequent report (Martin and Ghez 1993).