The development of effective biomarkers for detecting the magnitude of radiation exposure and resiliency of host response is crucial to identifying appropriate treatment strategies after radiation exposure. We hypothesized that the gastrointestinal resident bacteria would demonstrate predictable, dose-dependent changes after radiation exposure across two large animal models of acute radiation syndrome. Here, Göttingen minipigs (GMP) (n = 50) and rhesus macaques (n = 48) were exposed to five dose levels (resulting in mortality rates of 33-100% and 25-68.7%, respectively). Fecal samples taken prior to and after irradiation (day 0 for GMP; day 0, 3 and 14 for macaques) were used for 16S rRNA gene sequence amplicon high-throughput sequencing. Baseline gut microbiota profiles were dissimilar between GMP and macaques, however, radiation appeared to have similar effect at the phylum level, resulting in Bacteroidetes decrease and Firmicutes increase in both models. The abundance of the main Bacteroidetes genus ( Bacteroides for GMP, Prevotella for macaques) was profoundly decreased by irradiation. Intracellular symbionts [Elusimicrobia in GMP, Treponema (Spirochaetes) in macaques] consistently increased after irradiation, suggesting their use as potential biomarkers of intestinal injury, and potential negative effect on health. Prevotella, Lactobacillus, Clostridium XIVa, Oscillibacter and Elusimicrobium/ Treponema abundances were found to be very significantly correlated with radiation intensity. Furthermore, Prevotella, Enterorhabdus and Ruminococcus and Enterorhabdus maintenance was strongly associated with survival in GMP, while Prevotella, Oscillibacter and Treponema were strongly associated with survival and Streptococcus with death in macaques. Overall, we found that a wide range of gut bacterial genera known to be abundant in the human gut microbiota are excellent biomarkers of radiation intensity and resilience in animal models, and that detrimental effects can be monitored, and potentially prevented, by targeting selected genera.