A canonical understanding of how the brain controls arm movements assumes that each hemisphere delivers commands exclusively to the contralateral limb. This has been supported by a rich history of work in clinical neurology and experimental neuroscience. However, a growing body of research suggests that the ipsilateral hemisphere has some form of involvement as well. For example, neural activity recorded from a single hemisphere of the motor cortex may be used to accurately decode movements of the ipsilateral arm. The precise functions of these ipsilateral signals, and the bilateral nature of computations in the motor cortex more broadly, are poorly understood. In this thesis, we investigate the computations underlying isolated control of a single arm and coordinated control of both arms. Single-unit activity was recorded from both hemispheres of the motor cortex in non-human primates performing a range of behavioral tasks designed to test specific hypotheses regarding functional laterality. Moving our limbs requires a dynamic process of observing the environment we will interact with, selecting an appropriate action, preparing that action with the correct effector, and executing it. In Chapter 2, we ask: How does emerging population activity organize to form arm-specific signals as motor plans are prepared and executed? We find that the population signals are marked by two different components, which we refer to as “dedicated” and “distributed”. Dedicated signals were comprised of activity that was largely segregated for the two arms at the level of individual units, with activity most prominently located in the contralateral hemisphere. This component gradually emerged across preparation and movement and drove population activity for the two arms into divergent neural subspaces. In contrast, the distributed component represented a fundamentally bilateral function, as it contained behaviorally-specific information for both arms but did not segregate the population signals for each arm into separate neural subspaces. These two components allow for both independence and interaction of bilateral arm signals. Although nominally “motor” the motor cortex displays sensory responses as well. In Chapter 3, we ask: Is the postural state of both arms integrated into commands for a single arm? Using a task that manipulates the static posture of the stationary hand during unimanual reaching, we found that the relationship between neural activity and behavior was sensitive to the state of the stationary hand. Coding of the reach targets changed congruently with the posture of the stationary hand, i.e. the mapping between neural activity and reaching behavior shifted in the same direction that the posture of the stationary hand was moved. Our results offer mixed support for the intriguing hypothesis that ipsilateral activity reflects a parallel plan for how the stationary hand would move if it were selected for unimanual action. Alternatively, they may be interpreted in the context of bimanual coordination or reflect whole-body control that changes when the mechanical properties of the linked system are altered. Finally, in Chapter 4, we ask: Do network dynamics spanning the two hemispheres change flexibly to meet the coordination requirements of different bimanual tasks? Daily usage of our hands typically requires the two limbs to work in concert with one another, adopting task-specific coordination patterns. Here, we introduce a novel bimanual task that differentiates not only bimanual control from unimanual control, but independent bimanual control from bimanual coordination. We find that the mapping between neural activity and reaching behavior is altered between isolated unimanual movements and coordinated bimanual movements. However, the dynamics governing the evolution of neural activity appear to be largely maintained across control conditions. One interpretation of these results is that activity in the motor cortex coordinates commands for the two limbs not by adopting new interhemispheric dynamics, but by occupying different regimes of a context-invariant dynamical landscape. Our results regarding these questions offer new insights into the basic science of bilateral motor control and provide important foundations for future neurotechnologies like bimanual neuroprosthetics.