Smooth and efficient walking requires the coordination of movement across different parts of the body. The cerebellum plays an important role in this process, yet the specific neural circuit mechanisms of whole-body coordination are poorly understood. Although sophisticated genetic tools exist to manipulate the cerebellar circuit in mice, analyses of mouse gait have typically been limited to gross performance measures and lack detail about precision and timing of limb movements. In this project, I developed an automated, high-throughput, markerless 3D tracking system (LocoMouse) for quantifying locomotion in freely walking mice. Using LocoMouse, I showed that locomotor parameters for individual limbs vary systematically with mouse walking speed and body size. In visibly ataxic Purkinje cell degeneration (pcd) and reeler mice, I found that 3D limb trajectories and, especially, interlimb and whole-body coordination are specifically impaired. Our findings suggest a failure to predict the consequences of movement across joints, limbs, and body. These experiments were essential to establish a quantitative framework for whole-body locomotor coordination in mice (Machado, Darmohray et al. eLife 2015). The LocoMouse system was then combined with optogenetic tools to ask how different output regions of the cerebellum differentially contribute to locomotor coordination. I expressed ChR2 in Purkinje cells and stimulated their terminals in the medial, interposed, and lateral cerebellar nuclei of freely walking mice. Here, I identified locomotor parameters that were specifically related to the manipulation of each nucleus. Acute disruption of neural activity in medial and interposed nuclei immediately perturbed ongoing locomotion. In contrast, similar manipulation of Purkinje cell inputs to the lateral nucleus had no observable effect on ongoing locomotor behavior. These results are broadly consistent with previous anatomical and lesion studies suggesting a medial-to-lateral functional organization of cerebellar outputs. Taken together, these experiments isolated impairments in interlimb and whole-body coordination in mice with cerebellar manipulations. In contrast, spinal cord mutant mice revealed impairments at the intralimb level with no alteration in the interlimb coordination. I characterized distinct motor deficits associated with manipulations in different brain regions and identified and quantified core features of cerebellar ataxia in mice. These experiments establish the LocoMouse system, combined with genetic manipulations, as a powerful system to dissect cerebellar circuit mechanisms of coordinated locomotion.