As wolves recolonize across their former range in western North America, encounters between livestock and wolves are expected to increase in frequency. Understanding the physiological state of the prey, as a response to stress imposed by the presence of a predator (trait-mediated effects), will help with predicting the total effect of predators on their prey beyond direct consumption (density-mediated effects). Fecal glucocorticoid (GCM) is widely used to measure stress response, but provides inconclusive results, particularly when applied to a finer spatio-temporal variation in predation risk (i.e., wolf-livestock encounters). Since the impact of external stressors (i.e., wolf encounters) on the body influences ultimately the physical and metabolic state of the animal, we purpose to investigate: 1) if the fecal metabolome extracted from cattle fecal samples reflects changes in the GCM levels, and 2) if the cattle metabolome can better predict stress response than GCM post-wolf encounter. We first performed a controlled stress experiment on five captive cows in a pasture. We also conducted a field study in Washington where we fit GPS collars equipped with proximity sensors on two wolves in two separate packs, and on 40 range cows in four different livestock herds. When a wolf and a cow equipped with proximity sensors are within 128 meters from one another, the GPS collars send a real-time message that allows us to assess the physiological state of the prey at a finer spatial-temporal scale. In both studies, we collected fecal samples from GPS radio-collared cattle and compared their metabolome to GCM levels. Our results clearly show that the metabolomic analysis of the cattle gut microbiome can better predict acute stress response than the GCM, in both the controlled and free-ranging environment.