BackgroundThe fractal scaling evident in the step-to-step fluctuations of stepping-related time series reflects, to some degree, neuromotor noise. Research questionThe primary purpose of this study was to determine the extent to which the fractal scaling of step width, step width and step width variability are affected by performance of an attention-demanding task. We hypothesized that the attention-demanding task would shift the structure of the step width time series toward white, uncorrelated noise. MethodsSubjects performed two 10-min treadmill walking trials, a control trial of undisturbed walking and a trial during which they performed a mental arithmetic/texting task. Motion capture data was converted to step width time series, the fractal scaling of which were determined from their power spectra. ResultsFractal scaling decreased by 22% during the texting condition (p < 0.001) supporting the hypothesized shift toward white uncorrelated noise. Step width and step width variability increased 19% and five percent, respectively (p < 0.001). However, a stepwise discriminant analysis to which all three variables were input revealed that the control and dual task conditions were discriminated only by step width fractal scaling. SignificanceThe change of the fractal scaling of step width is consistent with increased cognitive demand and suggests a transition in the characteristics of the signal noise. This may reflect an important advance toward the understanding of the manner in which neuromotor noise contributes to some types of falls. However, further investigation of the repeatability of the results, the sensitivity of the results to progressive increases in cognitive load imposed by attention-demanding tasks, and the extent to which the results can be generalized to the gait of older adults seems warranted.