To walk in the real world, we continually alter our gait to cope with changing terrains, goals, and constraints on the body. This is a nontrivial feat—individual muscle activities must be adjusted to produce a desired gait and that desired gait must be selected from a myriad of possible coordination patterns. How this is accomplished is poorly understood. One principle that could guide the control of legged locomotion is an optimization process that seeks to meet some objective. The main goal of my thesis is to investigate the role of optimization, and in particular energy optimization, in the control of human gait. To do so, I undertook four distinct studies. In the first study, I developed a control system for a lower limb exoskeleton that leverages the body’s internal control by tapping directly into the user’s muscle activity. The myoelectric controller accurately identifies the user’s desired motion, automatically gradates the actuation of an exoskeleton, and is adaptable to varying gaits and terrains. In my second study, I sought to understand how entire coordination patterns, rather than individual muscle activities, may be optimized. I hypothesized that humans continuously optimize their gait to minimize energy expenditure. To test this, I used exoskeletons to alter the energetic consequences of various gaits. I made abnormal ways of walking energetically optimal and found that when given broad experience with the novel energetic landscapes subjects discovered the optimal gaits and opted to walk at them, even when the energetic benefits were small. In a third study, I found that the nervous system can be primed to initiate this optimization when perturbed toward low cost gaits, or can spontaneously initiate optimization when natural gait variability is high enough to elucidate a clear energetic gradient. Once optimization is initiated, I found evidence that the nervous system employs a ‘local search’ strategy to gradually descend energetic gradients and converge on novel optima. Given that energetic cost plays a central role in continuously shaping movement, in my fourth and final study I developed a technique to estimate instantaneous muscle energy use during non-steady state walking. This technique now makes it possible to measure a physiological signal that likely plays a central role in the control and optimization of human movement.