Anecdotally arm movements appear to be an intrinsic component of strategies that serve to maintain upright posture. The importance for such arm movements appears to be heightened in situations of balance perturbation or uncertainty. The purpose of this study was to investigate the role of arm movements whilst performing a moderately difficult balance task. Twenty-four healthy subjects (18 male; mean 34 years old) were asked to maintain their balance when adopting the Romberg position upon a wooden beam with a width of 85mm and a height of 50 mm. The beam was placed on a block of foam (with a width of 270 mm and a height of 85mm) thereby creating balance uncertainty. Subjects were requested to look ahead with their eyes open and use the handrail surrounding them only when their balance was compromised. The task was performed in 3 trials lasting 60s each, with two conditions (balanced order). Subjects were instructed to either cross (FIXED) or to use their arms (FREE) in order to maintain balance upon the beam. Mean (± SEM) measures of medio-lateral (M-L) and anterior-posterior (A-P) body sway were obtained using a force platform (Amt Inc. USA) and at the level of C7 and the head using Fastrak (Polhemus Inc. USA). Students paired t-tests were performed between population mean sway parameters for each trial and averaged over trials in each condition. The number of touches made by the subject onto the handrail was recorded as task failure. We found a significant reduction of mean (3 trials) sway path length in the M-L (2859 ± 345 to 2353 ± 198mm) and A-P (2353 ± 197 to 2089 ± 191mm) directions with the FREE condition. Similar trends were observed at the level of C7 and the head, although only the latter in the A-P (1831 ± 353 to 1440 ± 220mm) direction obtained significance. However, the effect of arm movement was predominantly manifest within the first trial – irrespective of parameter. Indeed, sway parameters converged by trial 3 presumably as a result of task adaptation. Further evidence for such a process is gleaned by the fact that task failure becomes less frequent in both conditions, although the rate is consistently reduced in the FREE condition (table 1.). In conclusion, free arm movements improve overall measures of stability, and perhaps more importantly significantly reduce task failure and speed up balance task adaptation compared to when arms were restricted.