Avoidance and tolerance are the two means by which plants cope with herbivores. Avoidances internal to the plant, such as morphology, chemical repellants, thorns, etc., have received considerable attention in the plant-herbivore literature, but relatively little consideration has been given to avoidances external to the plant. We develop a conceptual framework of external plant avoidances of herbivory based on foraging selection impedances (associational avoidances), behavioral impedances (indirect avoidances), and physical impedances (refuges) organized along axes of efficiency, degree of protection, and necessity of tolerance characteristics. Associational avoidances are uncommon for terrestrial mammalian herbivores compared to plant-insect or marine situations. Indirect avoidances mediated through herbivore territoriality, predator avoidance, and other behaviors independent of foraging decisions are probably common in nature, but few have been formally documented. Biotic and geologic refuges providing a physical impedance are the only avoidances shown to have implications for plant biodiversity. This is particularly true for geologic refuges, where there is not a tradeoff between competition and the refuge effect. Small geologic refuges (rock outcrops, cliffs, etc.) are more likely to also positively or negatively alter associated plant microenvironments than large geologic refuges (mesas, islands, etc.). In a survey, 86% of small refuge studies reported positive effects on plant diversity compared to 50% for large refuges. Geologic refuges in more productive environments were more important in protecting diversity than refuges in less productive, semiarid environments, and the effects of protection were greater in communities with short compared to long evolutionary histories of grazing. Other characteristics of refuges such as extent across the landscape and the manner they alter or ameliorate the environment, as well as characteristics of the herbivore such as small or large, generalist or specialist may also determine the effectiveness of refuges, but there are too few studies to assess these factors.