Background: Balance problems are common in cerebral palsy (CP) but etiology is often uncertain. The classic Romberg test compares ability to maintain standing with eyes open versus closed. Marked instability without vision is a positive test and generally indicates proprioceptive loss. From previous work showing diminished hip joint proprioception in CP, we hypothesized that static and dynamic balance without vision (positive Romberg) would be compromised in CP., Methods: Force plate sway and gait velocity data were collected using 3D motion capture on 52 participants, 19 with diplegic CP, 13 with hemiplegic CP, and 20 without disability. Center of mass (COM) and center or pressure (COP) velocity, excursion, and differences between COM and COP in AP and ML directions were computed from static standing trials with eyes open and closed. Mean gait velocity with and without dribble glasses was compared. Hip joint proprioception was quantified as the root mean square of magnitude of limb positioning errors during a hip rotation task with and without view of the limb. Mixed model repeated measures analysis of variance (ANOVA) was performed with condition as within-subject (EO, EC) and group as between-subject factors (hemiplegia, diplegia, controls). Sway characteristics and gait speed were correlated with proprioception values., Results: Groups with CP had greater sway in standing with eyes open indicating that they had poorer balance than controls, with the deficit relatively greater in the ML compared to AP direction. Contrary to our hypothesis, the decrement with eyes closed did not differ from controls (negative Romberg); however, proprioception error was related to sway parameters particularly for the non-dominant leg. Gait speed was related to proprioception values such that those with worse proprioception tended to walk more slowly., Conclusions: Postural instability is present even in those with mild CP and is yet another manifestation of their motor control disorder, the specific etiology of which may vary across individuals in this heterogeneous diagnostic category.