PURPOSE: The aim of this study was to determine if the linear and nonlinear components of the energy expenditure-walking speed relationship are influenced by body mass index (BMI; kg/m2). The secondary aims were to determine if the relationship was influenced by age, height, and sex. METHODS: Subjects (n=182) walked at 2, 3, and 4 mph for six minutes each with oxygen consumption (V̇O2; ml/kg/min) and measured via indirect calorimetry and converted to energy expenditure (EE; W/kg). Because of the curvilinear change in metabolic rate with increase in walking speed, polynomial random coefficient regression (PRCR) was employed to produce a model which captures the slope of change. Individual level linear and quadratic coefficients were analyzed for relationships with BMI, age, height, and sex. RESULTS: The net V̇O2 regression formula for walking was 1.79(x-3)2+4.97(x-3)+9.32 where x is speed in mph. BMI was modestly correlated with the quadratic coefficients (r = 0.15 to 0.17, p = 0.02 to 0.04) but not the linear coefficients (r =0.02- 0.07, p = 0.36-0.78) for V̇O2 and EE. There was no difference in coefficients between normal BMI (18.5-<25.0 kg/m2), overweight (25-<30.0 kg/m2) and obese (>30.0 kg/m2) groups (H = 1.5-4.0, p = 0.13-0.48). Delta V̇O2 for 2-3 mph, 3-4 mph, and 2-4 mph were not correlated with BMI (r = -0.02 - 0.13, p = 0.11 - 0.41). Height was inversely correlated with the linear and quadratic coefficients (r = -0.32 to -0.14, p = 0.09). Age was not correlated to coefficients (r = -0.16 to 0.32, p = 0.06-0.44). The coefficients for sex were not different after controlling for height in ANCOVA (F(1,179)=0.3-2.9, p >0.09). Age was not correlated to coefficients (r = -0.16 to –0.32, p = 0.06-0.44). CONCLUSION: Although BMI had a modest relationship with the quadratic coefficient, it explained less than 3% of the variance in V̇O2 or EE. Combined with the absence of a delta V̇O2 or a linear component, BMI does not influence the energy expenditure-walking speed relationship. Height explained up to 9% of the variance in the coefficients and eliminated apparent sex differences. Age was not related to the coefficients.