Objective Many biomechanical studies consistently show that a broader pelvis increases the reaction forces and bending moments across the femoral shaft, increasing the energetic costs of unloaded locomotion. However, a biomechanical model does not provide the real amount of metabolic energy expended in walking. The aim of this study is to test the influence of pelvis breadth on locomotion cost and to evaluate the locomotion efficiency of extinct Pleistocene hominins. Material and Methods The current study measures in vivo the influence of pelvis width on the caloric cost of locomotion, integrating anthropometry, body composition and indirect calorimetry protocols in a sample of 46 subjects of both sexes. Results We show that a broader false pelvis is substantially more efficient for locomotion than a narrower one and that the influence of false pelvis width on the energetic cost is similar to the influence of leg length. Two models integrating body mass, femur length and bi‐iliac breadth are used to estimate the net and gross energetic costs of locomotion in a number of extinct hominins. The results presented here show that the locomotion of Homo was not energetically more efficient than that of Australopithecus and that the locomotion of extinct Homo species was not less efficient than that of modern Homo sapiens. Discussion The changes in the anatomy of the pelvis and lower limb observed with the appearance of Homo ergaster probably did not fully offset the increased expenditure resulting from a larger body mass. Moreover, the narrow pelvis in modern humans does not contribute to greater efficiency of locomotion.