BACKGROUND:People at early stages of multiple sclerosis have subtle balance problems that may affect gait stability. However, differences in methods of determining stability such as sensor type and placements, may lead to different results and affect their interpretation when comparing to controls and other studies. QUESTIONS:Do people with multiple sclerosis (PwMS) exhibit lower gait stability? Do location and type of data used to calculate stability metrics affect comparisons? METHODS:30 PwMS with no walking impairments as clinically measured and 15 healthy controls walked on a treadmill at 1.2 ms-1 while 3D acceleration data was obtained from sacrum, shoulder and cervical markers and from an accelerometer placed at the sacrum. The local divergence exponent was calculated for the four data sources. An ANOVA with group (multiple sclerosis and control) and data source as main factors was used to determine the effect of disease, data source and their interaction on stability metrics. RESULTS:PwMS walked with significantly less stability according to all sensors (no interaction). A significant effect of data source on stability was also found, indicating that the local divergence exponent derived from sacrum accelerometer was lower than that derived from the other 3 sensor locations. SIGNIFICANCE:PwMS with no evident gait impairments are less stable than healthy controls when walking on a treadmill. Although different data sources can be used to determine MS-related stability deterioration, a consensus about location and data source is needed. The local divergence exponent can be a useful measure of progression of gait instability at early stages of MS.