In rheumatoid arthritis (RA) it is believed that symptoms associated with the progression of the disease result in a reduction in the physical activity level of the patient. One of the key flaws of the research surrounding this hypothesis to date is the use of non-validated physical activity outcomes measures. In this study, an algorithm to estimate physical activity levels in patients as they perform a simulated protocol of typical activities of daily living using SHIMMER kinematic sensors, incorporating tri-axial gyroscopes and accelerometers, is proposed. The results are validated against simultaneously recorded energy expenditure data and the defined activity protocol and demonstrate that SHIMMER can be used to accurately estimate physical activity levels in RA populations. © 2011 IEEE.