Purpose: The current study compared output from the vertical axis of two models of ActiGraph accelerometers under free-living and laboratory conditions in persons with multiple sclerosis (MS) and matched controls. Method: The sample included 41 cases of MS and 41 controls matched by age, sex, height and weight. Participants concurrently wore 7164 and GT3X accelerometers for 6 days under free-living conditions, and undertook up to five, 6-minute periods of walking at five speeds on a motor-driven treadmill. Results: Under free-living conditions, there was substantial, but not absolute, agreement (Intraclass correlation = 0.983) between devices and the output from the model 7164 accelerometer was significantly greater (p < 0.01) than that of the GT3X, resulting in a 7.0% difference between devices. The treadmill-walking conditions indicated that this difference in output between accelerometers was based on slow-walking speeds where there was a statistically significant (p < 0.01) and 30.3% discrepancy. Conclusions: The ActiGraph model 7164 and GT3X accelerometers are not interchangeable under free-living conditions, and the discrepancy in measurement between devices might be explained by the difference in output under slow-walking conditions. Researchers and clinicians should be aware of the difference in output between generations of ActiGraph accelerometers, and perhaps other brands, in future applications involving persons with MS and controls. Implications for Rehabilitation There is a discrepancy in output between ActiGraph accelerometer models under free-living conditions and slow-treadmill walking. Researchers should consider developing equations to translate outputs between ActiGraph accelerometers. Clinicians should be aware that these accelerometers are not interchangeable within and between applications involving persons with multiple sclerosis. © 2013 Informa UK, Ltd.