OBJECTIVE:To determine the effects of aging on balance control during walking. DESIGN:Two-group repeated-measures design. SETTING:Gait laboratory in Australia. PARTICIPANTS:Convenience sample of 20 healthy older subjects (mean age, 72y) and 20 healthy young subjects (mean age, 24y). INTERVENTIONS:Changes in locomotor performance in response to perturbations to balance were quantified for healthy older adults compared with healthy young adults for (1) straight line walking at preferred speed, (2) straight line walking at fast speed, (3) figure-of-eight walking at preferred speed, and (4) figure-of-eight walking while performing a secondary motor task. MAIN OUTCOME MEASURES:Gait speed, stride length, cadence, and double-limb support duration, using a footswitch system. RESULTS:Healthy older people screened for pathology had gait patterns comparable to young adults for straight line walking at preferred speed. However, multivariate analysis of variance (MANOVA) showed a significant interaction between age and speed when balance was perturbed by requiring subjects to change from walking at preferred to fast speeds (Pillai-Bartlett trace=.259, F(4,35)=3.06, P<.029, partial eta(2)=.259). This occurred because older people did not increase their speed (F(1,38)=7.65, P<.01, partial eta(2)=.168) or stride length (F(1,38)=12.23, P<.01, partial eta(2)=.243) as much as did the young adults. MANOVAs did not show statistically significant interactions between age and turning conditions or age and dual task conditions, although older people walked more slowly and with shorter steps when turning or performing a secondary task. CONCLUSIONS:Balance strategies during gait are task specific and vary according to age. In response to challenges to balance imposed by the requirement to change from preferred to fast walking, older people did not increase their speed and stride length to the same extent as did younger adults. This was possibly a strategy to maintain their stability.