1 A field experiment involving exclosure techniques was used to assess the relative roles of arthropods, molluscs and rodent herbivores in the survival and growth of temperate grassland plants. The study focused on the impact of herbivores on plant survival, above-ground biomass and root weight ratios of 21 plant species which were sown experimentally in a grassland and meadow. 2 Of the three herbivore groups studied, rodents exerted the greatest influence on plant performance, reducing plant biomass by as much as 50% and substantially increasing plant mortality. Molluscs significantly decreased plant numbers but plants appeared more able to compensate for biomass lost through grazing. By comparison, arthropods played only a minor role in determining either plant biomass or survival. 3 Positive correlations in the responses of plant species to herbivory by molluscs and rodents indicate that the major difference in the impact of these two guilds of herbivores on plant performance (survival, biomass and root weight ratio) is in the magnitude rather than the direction of the plant response. Legumes were more susceptible to herbivory than grasses, exhibiting both lower survival and greater loss of biomass. 4 Results indicate that where resources are limiting, as in the meadow, the direct influence of herbivores on plant numbers is minimal since herbivore induced mortality is only one component of naturally low plant survival. In contrast, the influence of herbivores on plant growth can be considerable since plants are often unable to compensate for tissue loss via regrowth. Thus, in the grassland, the overall influence of herbivores was on plant numbers whereas in the meadow their main impact was on mean plant biomass. 5 Although the majority of the variation in both plant survival and biomass occurred independently of the experimental treatments, with almost 50% attributable to intrinsic species effects, indirect evidence reveals herbivores to have an impact on plant community composition. Plants which suffer both high mortality and poor growth as a result of herbivory (e.g. Trifolium repens, T. pratense) are rare while those plants whose performance is little influenced by herbivores (e.g. Agrostis capillaris, Dactylis glomerata, Plantago lanceolata) are abundant in the grassland communities studied.