Grazing distribution patterns of large herbivores are affected by abiotic factors such as slope and distance to water and by biotic factors such as forage quantity and quality. Abiotic factors are the primary determinants of large-scale distribution patterns and act as constraints within which mechanisms involving biotic factors operate. Usually there is a proportional relationship between the time large herbivores spend in a plant community and the available quantity and quality of forage. This grazing pattern may result from decisions made by animals at different spatial and temporal scales. Foraging velocity decreases and intake rate increases in areas of abundant palatable forage. These non-cognitive mechanisms that occur at smaller spatial scales (bites, feeding stations, small patches) could result in observed grazing patterns. However, large herbivores also appear to select areas (patches and feeding sites) to graze. Optimal foraging models and other models assume animals use "rules of thumb" to decide where to forage. A cognitive mechanism assumes animals use spatial memory in their foraging decisions. With such abilities, large herbivores could return to nutrient-rich sites more frequently than to nutrient-poor sites. Empirical studies indicate that large herbivores have accurate spatial memories and have the ability to use spatial memory to improve foraging efficiency. Body size and perceptual abilities can constrain the choices animals can make during foraging. A conceptual model was developed to demonstrate how cognitive foraging mechanisms could work within constraints imposed by abiotic factors. Preliminary predictions of the model correspond to observed grazing patterns. Recognizing that large herbivores may use previous experiences to decide where to forage may be useful in developing new techniques to modify grazing patterns. Grazing distribution patterns appear to result from decisions and processes made at a variety of spatial and temporal scales.