The annual plant Arabidopsis thaliana is widely used as a model system in molecular genetics, but little is known about populations in the field. In this experimental field study of natural populations of Arabidopsis, I tested the assumption that plant resistance has fitness costs. Models of the evolution of resistance assume a cost, which is envisioned as a reduction in fitness in the absence of natural enemies, such as insect herbivores and pathogens. The presumed basis of this cost is the diversion of limiting resources away from present and future growth and reproduction. Recent failures to detect allocation costs of resistance to herbivores have raised questions about whether costs exist and, thus, about the appropriateness of theories that postulate such costs. I found genetic variation for two traits commonly thought to function as resistance characters: trichome density and total glucosinolate concentration. Under field conditions, these characters both reduced damage by the natural assemblage of herbivores and exhibited significant fitness costs.