Zooarchaeological data on small game use hold much potential for identifying and dating Paleolithic demographic pulses in time and space, such as those associated with modern human origins and the evolution of food-producing economies. Although small animals were important to human diets throughout the Middle, Upper, and Epi-Paleolithic periods in the Mediterranean Basin, the types of small prey emphasized by foragers shifted dramatically over the last 200,000 years. Slow-growing, slow-moving tortoises, and marine mollusks dominate the Middle Paleolithic record of small game exploitation. Later, agile, fast-maturing animals became increasingly important in human diets, first birds in the early Upper Paleolithic, and soon thereafter hares and rabbits. While the findings of this study are consistent with the main premise of Flannery's Broad Spectrum Revolution (BSR) hypothesis (Flannery, K. V. (1969). In Ucko, P. J., and Dimbleby, G. W. (eds.), The Domestication and Exploitation of Plants and Animals, Aldine Publishing Company, Chicago, pp. 73–100), it is now clear that human diet breadth began to expand much earlier than the Pleistocene/Holocene transition. Ranking small prey in terms of work of capture (in the absence of special harvesting tools) proves far more effective in this investigation of human diet breadth than taxonomy-based diversity analyses published previously. Our analyses expose a major shift in human predator–prey dynamics involving small game animals by 50–40 KYA in the Mediterranean Basin, with earliest population growth pulses occurring in the Levant. In a separate application to the Natufian period (13–10 KYA), just prior to the rise of Neolithic societies in the Levant, great intensification is apparent from small game use. This effect is most pronounced at the onset of this short culture period, and is followed by an episode of local depopulation during the Younger Dryas, without further changes in the nature of Natufian hunting adaptations. An essential feature of the diachronic and synchronic approaches outlined here is controlling the potentially conflating effects of spatial (biogeographic) and temporal variation in the faunal data sets.