We tested how lateral stability affects gait as a function of age. A simple computational model suggests that walking is laterally unstable and that age-related decreases in motor and sensory function may be treated as noise-like perturbations to the body. Step width variability may be affected by active control of foot placement subject to noise. We hypothesized that age-related deficits may lead to increased step width variability. A possible compensation would be to walk with wider steps to reduce the lateral instability. The addition of external stabilization, through elastic cords acting laterally on the body during treadmill walking, would be expected to yield reduced step width variability and/or reduced average step width. We measured step width, its variability (defined as standard deviation), and metabolic energy expenditure in eight adult human subjects aged less than 30 years (Young) and ten subjects aged at least 65 years (Old). Subjects walked with and without external stabilization, each at a self-selected step width as well as a prescribed step width of zero. In normal walking, Old subjects preferred 41% wider steps than Young, and expended 26% more net energy (P < 0.05). External stabilization caused both groups to prefer 58% narrower steps. In the prescribed zero step width condition, Old subjects walked with 52% more step width variability and at 20% higher energetic cost. External stabilization resulted in reduced step width variability and 16% decreased energetic cost. Although there was no significant statistical interaction between age group and stabilization, Old and Young subjects walked with similar energetic costs in the stabilized, prescribed step width condition. Age-related changes appear to affect lateral balance, and the resulting compensations explain much of the increased energetic cost of walking in older adults.