Variability is a natural and important feature of human movement. Using existing theoretical frameworks as a foundation, we propose a new model to explain movement variability as it relates to motor learning and health. We contend that mature motor skills and healthy states are associated with an optimal amount of movement variability. This variability also has form and is characterized by a chaotic structure. Less than optimal movement variability characterizes biological systems that are overly rigid and unchanging, whereas greater than optimal variability characterizes systems that are noisy and unstable. Both situations characterize systems that are less adaptable to perturbations, such as those associated with abnormal motor development or unhealthy states. From our perspective, the goal of neurologic physical therapy should be to foster the development of this optimal amount of movement variability by incorporating a rich repertoire of movement strategies. The development of such a repertoire can be enhanced by incorporating a multitude of experiences within the therapeutic milieu. Promoting complex variation in human movement allows either motor development or the recovery of function after injury not to be hard coded, but determined instead by the active engagement of the individual within their environment. Measurement tools derived from nonlinear dynamics that characterize the complexity of movement variability provide useful means of testing these propositions. To illustrate, we present 2 clinical case studies, one pediatric and one adult, where we applied our theoretical framework to measuring change in postural control.