The ratio of the upper to lower limb lengths [or the intermembral index (IMI)] in the earliest human ancestors is closer to that of the living chimpanzees than to our own, although the former show undoubted adaptations to bipedality. What biomechanical factors could then have led to the phenomenon of genus Homo? This paper proposes and evaluates a relationship between IMI and hand-carrying. Assuming that coordination of limb swing frequencies of the upper and lower limbs would be the subject of positive selection, a mathematical expression was derived and can in part explain the changes in IMI. We found that AL-288-1 [3.6 million years old (MY)], the most complete skeleton of the early hominid Australopithecus afarensis, could only have carried loads equivalent to 15-50% of the upper limb weight while maintaining swing symmetry, but KNM WT-15000, Homo ergaster (1.8MY) and modern humans could both carry loads 3 times heavier than the upper limb while maintaining swing symmetry. The carrying ability of chimpanzees would be inferior to that of AL-288-1. The IMI of modern humans, at 68-70, is the smallest, and is optimal for hand-carrying under our criteria. Under reduced selection pressure for hand-carrying, but unreduced selection for mechanical effectiveness, we might expect humans to evolve a longer upper limb, to improve swing symmetry when unloaded.