The formation of mixed-species groups (MSGs) (also called heterospecific or polyspecific groups) is often considered an anti-predator strategy used in response to an increased predation risk. Recent studies performed in savanna ecosystems in Africa have demonstrated an increased formation of ungulate MSGs in higher predation risk areas. Nevertheless, these studies only considered one predator species (the lion Panthera leo) in evaluating the response to predation risk. Yet, species evolve in multi-predator systems that require prey to respond to multiple sources of risks. Although the importance of incorporating these multiple predators in studies is recognized, there is still little information on the behavioral strategies, including MSG formation, that prey can use to minimize predation risk from multiple predators. In the present study, we analyzed a large dataset of camera trap data (more than 160,000 observations) from five sites located in savanna biome in South Africa. We studied the likelihood of MSGs formation in response to the presence of different predators (lion, leopard (Panthera pardus), spotted hyena (Crocuta crocuta), and wild dog (Lycaon pictus)). We demonstrated that the probability of forming MSGs increased when (i) the photographic rate of the interacting species increased and (ii) the photographic rate of predators increased. Our results showed that lion and spotted hyena are the predator species most likely to trigger the formation of MSGs for different prey species. These results suggest a “hierarchy of fear” among predators in which both species are the most feared, even for prey species that either predator does not preferentially select. A large number of MSGs formed in the presence of these two predator species included Burchell’s zebras (Equus quagga burchellii), in association with species that have complementary modes of predator detection (e.g. wildebeest (Connochaetes taurinus) and giraffe (Giraffa camelopardalis). Our study highlights the importance of incorporating multiple predators into MSG analyses to better understand ecosystem functioning and, community formation.