PURPOSE: The purpose of this study was to evaluate the validity of two published accelerometer analysis methods to estimate energy cost of physical activities in adults with and without functional limitations. METHODS: 62 healthy and functionally limited ambulatory adults underwent measures of strength and functional testing to cluster into 1) no functional limitations [HLTHY], or 2) lower and/or upper body limitations [LIM]. All subjects completed a variety of simulated activities while wearing accelerometers and a portable metabolic system. Estimated METs were derived from a hip-worn accelerometer (Freedson et al.(HIP)), and a wrist-worn accelerometer (Staudenmayer et al.(WRIST)). Model predictions were compared to measured METs and evaluated on the types of activities for which they were developed (HIP=locomotion, WRIST=all activities). Statistical significance was assessed with a linear mixed model. RESULTS: Mean (95% CI) measured METs across all activities were HLTHY (n=17, 48.1 ± 16.7 yrs) 3.17 (1.59) and LIM (n=45, 61.0 ±12.9 yrs) 2.32 (1.08). Fig.1. Accelerometer Model Prediction Performance. Mean (95% CI). Fig.1. shows neither the HIP nor the WRIST methods are significantly biased for the HLTHY group, and both methods are biased for the LIM group. Estimates for the LIM group are significantly different from the HLTHY group for both methods. The HIP and WRIST methods under- and over-estimate respectively for the LIM group. CONCLUSION: Accelerometer energy cost model performance is influenced by whether or not an individual has functional limitations. Future research efforts should examine how to improve accelerometer model performance within individuals with functional limitations. Supported by NIH-1R21HD080828.