Changes in lunar illumination alter the balance of risks and opportunities for animals at night, influencing activity patterns and species interactions. Our knowledge about behavioral responses to moonlight is incomplete, yet it can serve to assess and predict how species respond to environmental changes such as light pollution or loss of canopy cover. As a baseline, we wish to examine if and how wildlife responds to the lunar cycle in some of the darkest places inhabited by terrestrial mammals: the floors of tropical forests.We quantified the prevalence and direction of activity responses to the moon in tropical forest mammal communities. Using custom Bayesian multinomial logistic regression models, we analyzed long-term camera trapping data on 88 mammal species from 17 protected forests on three continents. We also tested the hypothesis that nocturnal species are more prone to avoiding moonlight, as well as quantified diel activity shifts in response to moonlight.We found that, apparent avoidance of moonlight (lunar phobia, 16% of species) is more common than apparent attraction (lunar philia, 3% of species). The three species exhibiting lunar philia followed diurnal or diurnal-crepuscular activity patterns. Lunar phobia, detected in 14 species, is more pronounced with higher degree of nocturnality, and is disproportionately common among rodents. Strongly lunar phobic species were less active during moonlit nights, which in most cases also decreases their total daily activity.Our findings indicate that moonlight influences animal behavior even beneath the forest canopy. This suggests that such impacts may be exacerbated in degraded and fragmented forests. Additionally, the effect of artificial light on wild communities is becoming increasingly apparent. Our study offers empirical data from protected tropical forests as a baseline for comparison with more disturbed areas, together with a robust approach for detecting activity shifts in response to environmental change.Open Research statement:The data and code for performing the analyses described in this article are available athttps://github.com/richbi/TropicalMoon.