In persons with multiple sclerosis (PwMS), synchronizing walking to auditory stimuli such as to music and metronomes have been shown to be feasible, and positive clinical effects have been reported on step frequency and perception of fatigue. Yet, the dynamic interaction during the process of synchronization, such as the coupling of the steps to the beat intervals in music and metronomes, and at different tempi remain unknown. Understanding these interactions are clinically relevant, as it reflects the pattern of step intervals over time, known as gait dynamics. 28 PwMS and 29 healthy controls were instructed to walk to music and metronomes at 6 tempi (0-10% in increments of 2%). Detrended fluctuation analysis was applied to calculate the fractal statistical properties of the gait time-series to quantify gait dynamics by the outcome measure alpha. The results showed no group differences, but significantly higher alpha when walking to music compared to metronomes, and when walking to both stimuli at tempi + 8, + 10% compared to lower tempi. These observations suggest that the precision and adaptation gain differ during the coupling of the steps to beats in music compared to metronomes (continuous compared to discrete auditory structures) and at different tempi (different inter-beat-intervals). The study of temporal correlations in step or stride intervals over time-also known as gait dynamics 1-provides useful insights on the neural control of locomotion in young adults 2,3 , healthy older adults 4 , and patients with movement disorders such as Parkinson's disease 5 , Huntington's disease 5,6 or multiple sclerosis (MS) 7. Gait gives rise to non-stationary inter-step/stride-interval signals, with temporal fluctuations which can be analyzed via non-linear methods 1. An example of these methods, capable of capturing the complexity of time-evolving behavior in the domain of gait analysis, is detrended fluctuation analysis (DFA). This analysis method can be applied to a time series obtained from gait measurements such as inter-step/stride-intervals 2,3,8. DFA is robust to non-stationaries in the data, often observed in gait interval time series. This method scales the long-term auto-correlations of non-stationary signals and quantifies the fluctuations in the time series using its self-similar property 8,9 with a value of fractal scaling index 'alpha' 8,9. Alpha provides an estimation of statistical 'persistence' or 'anti-persistence' in a time series 1. A healthy value of alpha in gait is between 0.5 and 1.0 (1 being highly persistent), and indicates the presence of statistical persistence within the inter-step-intervals 1,10. This means that the inter-step-intervals between consecutive steps are non-random and constant at a long range, with small deviations still being present across multiple consecutive strides. On the other hand, a value of alpha < 0.5 signifies