The sequence in which nonhuman carnivores consumed the edible parts of 260 larger herbivore carcasses in the Serengeti region of Tanzania proceeds from (1) hindquarter flesh, to (2) ribcage and forequarter flesh, (3) head flesh, (4) hindlimb marrow, (5) forelimb marrow, and lastly, (6) head contents. Only minor deviations from the regional consumption sequence occur, which are based on energetic and mechanical factors related to differences in carcass size and age, and consumer type. The results are used to model the archaeological signatures of scavenging and hunting. Assemblages accumulated by scavenging are expected to show a positive relationship between skeletal part representation and the frequencies by which the parts survive consumption by modern carnivores with edible tissues. Assemblages accumulated by hunting should show a negative relationship between the variables. Variations in the magnitude of the predicted relationships will be effected by carcass size and local degrees of competition for carcasses, but in a similar manner for scavenging and hunting. The efficacy of carcass consumption sequences for distinguishing hunting from scavenging archaeologically is compared to the carcass disarticulation sequence used by Potts (1983) , and bone survivorship data used by Binford (1984) . A preliminary application of the consumption sequence model to the Klasies River Mouth fauna indicates that medium-sized (size 3) and large (size 4) bovids at Klasies were scavenged primarily. Smaller (size 1 and 2) bovids show a distinct pattern, one more consistent with hunting. These results are similar to Binford's (1984) , but for reasons that are argued to have a firmer empirical basis.