The distribution of organisms across the globe forms non-random patterns that are based on their selection for certain environmental features. Habitat selection is the study of how organisms choose where they live. It is a fundamental behavior shaping several ecological and evolutionary processes. The hierarchical nature of the environment makes it imperative to consider multiple scales in habitat selection studies. Even though there is extensive coverage of scaling in habitat selection literature, most published papers only consider one scale. To our knowledge, habitat selection studies of the Antillean manatee (Trichechus manatus manatus) have never included more than one spatial scale. Hence, we modeled Antillean manatee habitat selection in the Sian Ka'an Biosphere Reserve, in Mexico, at two spatial scales: study area and 1 km buffer. We used GPS coordinates of opportunistic encounters (n = 102) recorded since 2009. We randomly generated 500 pseudo-absences per presence point for both scales, and extracted the environmental conditions from each point: seagrass abundance (leaf area index or LAI), water depth, and shortest distance to land, closest river and closest resting hole. We fitted a binomial regression of the probability of presence as a function of the environmental parameters using a Bayesian approach. Our results show that the probability of manatee presence increases in proximity to resting holes for both scales and increases with land proximity at the large-scale. Overall, we showed that proximity to resting holes is the most important feature of habitat selection for Antillean manatees in Sian Ka'an, followed by proximity to land. This study demonstrates the importance of multi-scale designs in habitat selection and highlights the need for more studies looking at the use and ecological implications of manatee resting holes.