We tested whether the rate at which force is applied to the ground sets metabolic rates during classical-style roller skiing in four ways: 1) by increasing speed (from 2.5 to 4.5 m/s) during skiing with arms only, 2) by increasing speed (from 2.5 to 4.5 m/s) during skiing with legs only, 3) by changing stride rate (from 25 to 75 strides/min) at each of three speeds (3.0, 3.5, and 4.0 m/s) during skiing with legs only, and 4) by skiing with arms and legs together at three speeds (2.0-3.2 m/s, 1.5° incline). We determined net metabolic rates from rates of O₂ consumption (gross O₂ consumption  standing O₂ consumption) and rates of force application from the inverse period of pole-ground contact [1/t_p(arms)] for the arms and the inverse period of propulsion [1/t_p(legs)] for the legs. During arm-and-leg skiing at different speeds, metabolic rates changed in direct proportion to rates of force application, while the net ground force to counteract friction and gravity () was constant. Consequently, metabolic rates were described by a simple equation (_metab =  · 1/t_p · C, where _metab is metabolic rates) with cost coefficients (C) of 8.2 and 0.16 J/N for arms and legs, respectively. Metabolic rates predicted from net ground forces and rates of force application during combined arm-and-leg skiing agreed with measured metabolic rates within ±3.5%. We conclude that rates of ground force application to support the weight of the body and overcome friction set the energetic cost of skiing and that the rate at which muscles expend metabolic energy during weight-bearing locomotion depends on the time course of their activation.