Recent discussion of the selective pressures leading to the evolution of modern human postcranial morphology, seen as early as Homo erectus, has focused on the relative importance of walking versus running. Specifically, these conversations have centered on which gait may have been used by early Homo to acquire prey. An element of the debate is the widespread belief that quadrupeds are constrained to run at optimally efficient speeds within each gait, whereas humans are equally efficient at all running speeds. The belief in the lack of optimal running speeds in humans is based, however, on a number of early studies with experimental designs inadequate for the purpose of evaluating optimality. Here we measured the energetic cost of human running (n=9) at six different speeds for five minutes at each speed, with careful replicates and controls. We then compared the fit of linear versus curvilinear models to the data within each subject. We found that individual humans do, in fact, have speeds at which running is significantly less costly than at other speeds (i.e., an optimal running speed). In addition, we demonstrate that the use of persistence hunting methods to gain access to prey at any running speed, even the optimum, would be extremely costly energetically, more so than a persistence hunt at optimal walking speed. We argue that neither extinct nor extant hominin populations are as flexible in the chosen speeds of persistence hunting pursuits as other researchers have suggested. Variations in the efficiency of human locomotion appear to be similar to those of terrestrial quadrupeds.