During the past decade, the idea that expertise is dead, or at best moribund, has become commonplace. Knowledge resistance appears to be increasing across a wide range of science-based topics, such as agriculture, evolution and genetics, vaccination, and climate change. But in many of these areas, denying expert authority is cost-free in everyday behavior, making it more rational for people to prize identity and group affiliation over realism. To probe the health of expertise in a domain with everyday consequences for knowledge resistance, we conducted three incentive-compatible studies of laypeople's preferences for sources of information they would read about specific medical conditions (e.g. heart disease, cancer, COVID-19). We found quite rational preference patterns, by which people preferred expert sources (physicians and scientists) over non-experts (celebrities and politicians) and group consensus (professional societies, polls) over individual opinions. These findings held most strongly for issues of personal medical concern, but were robust for less concerning health conditions, and for the highly politicized topic of COVID-19. Individuals who scored higher in intellectual humility and preferences for rational over experiential thinking were more likely to prefer the most expert sources.