The goal of this study was to examine the predictive value of sensory and motor system connectivity and lesion type on hand function in children with unilateral spastic cerebral palsy (USCP). In typically developing children, sensory and motor functions are primarily mediated via crossed (contralateral) connections, as opposed to same-sided (ipsilateral) connections. In children with USCP, an early brain injury often leads to sensory-motor reorganization, where ipsilateral motor connections on the lesioned side are seen to persist. Ipsilateral sensory connections are less common. Given the large scale adaptation in the motor system to unilateral brain injury and smaller scale adaptation in the sensory system, we hypothesized that loss of crossed sensory connections would be more predictive of hand function than loss of crossed motor connections. In addition, sensory circuits being established early in development has been shown to be more affected by cortical lesions than periventricular (PV) lesions. Thus, we hypothesized that cortical lesions would cause greater hand impairment than PV lesions. 23 children with USCP participated in anatomical, physiological and behavior testing of the sensory and motor systems. Anatomically, lesions were denned using MRI, and principal sensory and motor pathways were reconstructed using DTI. Physiologically, somatosensory evoked potentials (SSEPs) were recorded in response to vibrotactile stimulation of the index fingers; and EMG responses to motor cortex stimulation with TMS were recorded. For behavior, sensory hand function was tested with the Cooper stereognosis test and two-point discrimination; while motor hand function was assessed with the Jebsen-Taylor Test and box and blocks. We found that for the impaired hand, presence of significantly discriminable SSEP in the contralateral hemisphere, and a preserved contralateral sensory tract in DTI were both highly predictive of good hand motor and sensory function (Cohen’s kappa agreement between group of subjects with good hand function and testing modality = 0.83 and 0.48 respectively, with kappa > 0.75 being strong agreement, and kappa < 0.40 poor). Lesion type was also predictive of hand function (kappa = 0.47). By contrast, the presence of contralateral motor connections—both physiological from TMS and anatomical from DTI—was less predictive of hand function (kappa = 0.24 and 0.32 respectively). Thus, sensory connectivity is more predictive of hand function in children with USCP than motor connectivity. This indicates that therapies to restore anatomical and physiological sensory connectivity may be important for recovery of sensory-motor function.