Infected hosts may preserve fitness by resisting parasites (reducing parasite burden) and/or tolerating them (preventing or repairing infection-induced damage). Theory predicts that these individual-level defense strategies generate divergent population-level feedbacks that would maintain genetic heterogeneity for resistance but purge heterogeneity for tolerance. Because resistance reduces parasite abundance, selection for costly resistance traits will weaken as resistance becomes common. Such negative frequency-dependent selection contrasts with predictions for tolerance, which maintains parasite abundance and so is expected to generate positive frequency-dependent selection, unless, for example, tolerance trades off with resistance. Thus far, there have been few tests of this theory in natural systems. Here, we begin testing the predictions in a mammalian field system, using data on individual gastrointestinal nematode burdens, nematode-specific antibody titers (as a resistance metric), the slope of body weight on parasite burden (as a tolerance metric), and fitness from an unmanaged population of Soay sheep. We find that nematode resistance is costly to fitness and underpinned by genetic heterogeneity, and that resistance is independent of tolerance. Drawing upon empirical metrics such as developed here, future work will elucidate how resistance and tolerance feedbacks interact to generate population-scale patterns in the Soay sheep and other field systems.