The world’s first archaeological traces from 2.6 million years ago (Ma) at Gona, in Ethiopia, include sharp-edged cutting tools and cut-marked animal bones, which indicate consumption of skeletal muscle by early hominin butchers. From that point, evidence of hominin meat-eating becomes increasingly more common throughout the Pleistocene archaeological record. Thus, the substantive debate about hominin meat-eating now centers on mode(s) of carcass resource acquisition. Two prominent hypotheses suggest, alternatively, (1) that early Homo hunted ungulate prey by running them to physiological failure and then dispatching them, or (2) that early Homo was relegated to passively scavenging carcass residues abandoned by carnivore predators. Various paleontologically testable predictions can be formulated for both hypotheses. Here we test four predictions concerning age-frequency distributions for bovids that contributed carcass remains to the 1.8 Ma. old FLK 22 Zinjanthropus (FLK Zinj, Olduvai Gorge, Tanzania) fauna, which zooarchaeological and taphonomic data indicate was formed predominantly by early Homo. In all but one case, the bovid mortality data from FLK Zinj violate test predictions of the endurance running-hunting and passive scavenging hypotheses. When combined with other taphonomic data, these results falsify both hypotheses, and lead to the hypothesis that early Homo operated successfully as an ambush predator.