This study further validates accelerometers as a measure of walking impairment in persons with multiple sclerosis. We examined total movement counts and three novel methods of processing accelerometer data (i.e. standard deviation, approximate entropy and detrended fluctuation analysis) for quantifying real-life walking impairment in this population. A total of 70 individuals with a definite diagnosis of multiple sclerosis completed a battery of patient-rated measures of walking impairment and then wore an ActiGraph accelerometer for 7 days. The data were analyzed using multivariate analysis of variance and bivariate correlation analysis. The results indicated that total daily movement counts and standard deviation of daily movement counts differed between groups of persons with mild, moderate, and severe self-reported disability status and who were independently ambulatory or ambulatory with assistance. Those two metrics for the accelerometer data further demonstrated strong correlations with patient-rated measures of walking impairment. By comparison, there were smaller and often non-significant differences in approximate entropy and detrended fluctuation analysis metrics for the accelerometer data as a function of disability and ambulatory status, and only moderate correlations with patient-rated measures of walking impairment. The results confirm that the metric of total daily movement counts correlates with level of disability, ambulatory status, and patient reports of walking impairment in persons with multiple sclerosis. We further demonstrate that variability, indexed by the standard deviation of daily movement counts, correlates with multiple sclerosis-related disability, ambulatory status, and self-reported walking impairment. Such results provide preliminary evidence that variability in accelerometer counts is not simply noise and may provide important information about multiple sclerosis-related walking impairment.