The evident ataxic characteristics of gait in patients with cerebellar damage suggest that the cerebellum plays an important role in the neural control of gait. Ataxic features, such as increased gait variability and increased step width, are often related to gait stability. However, the link between these measures and gait stability is not straightforward. Therefore, to gain more insights into relations between gait stability, gait variability and gait ataxia, we quantified gait stability using the short-term maximum Lyapunov exponent. This is a more valid measure of gait stability, derived from dynamical systems theory. Eighteen patients with focal cerebellar lesions after tumor resection walked on an instrumented treadmill at 1.0m/s for 3min. The patients displayed relatively mild functional deficits (ICARS=6.9±6.4, range 0-20) and had a lower overground walking speed as compared to healthy controls (1.12m/s versus 1.31m/s). During treadmill walking, the short-term maximum Lyapunov exponent was higher in cerebellar patients, indicating reduced gait stability. Furthermore, step width was increased in the patient group while other spatio-temporal gait parameters were similar. Patients with the largest lesions in the vermis displayed the least stable gait pattern. These observations imply that the short-term maximum Lyapunov exponent is a sensitive measure of gait deficits in mildly ataxic cerebellar patients.