Ecologists that study the population dynamics of large and small herbivorous mammals operate in two worlds that overlap only partly, and in this paper I address whether the conjecture that these worlds represent two distinct and valid paradigms is currently justified. I argue that large mammals fall into three groups depending on whether they have effective predators or not, and whether they are harvested by humans. Because of human persecution of large predators, more and more large herbivorous mammals are effectively predator-free and are controlled bottom-up by food. But in less disturbed systems, large herbivorous mammals should be controlled top-down by effective predators, and this can lead to a trophic cascade. Small herbivorous mammals have been suggested to be controlled top-down by predators but some experimental evidence has challenged this idea and replaced it with the notion that predation is one of several factors that may affect rates of population increase. Intrinsic control (territoriality, infanticide, social inhibition of breeding) appears to be common in small herbivorous mammals with altricial young but is absent in species with precocial young, in ecosystems with strong stochastic weather variation (deserts) and in areas of human-induced habitat fragmentation in agricultural monocultures. The extrinsic control of large herbivores with precocial young validates part of Graeme Caughley’s Grand Vision, but the relative role of intrinsic and extrinsic mechanisms for small herbivores with altricial young is still controversial. An improved knowledge of population control mechanisms for large and small herbivores is essential for natural resource management.