The activity patterns of mammals depend on environmental changes (e.g., moon luminosity, food availability, weather) and endogenous rhythms. Behavioral observations are traditionally used to estimate the activity patterns of animals, but low visibility and the cryptic nature of some species entail that, in certain conditions, the animal is visible only for around 60% of the time. Recent advances in technology allow automatic data collection on the activity levels of animals. We used five years of data collected via accelerometers to understand how moon luminosity, seasonality, sex, and weather conditions influence the activity levels of the nocturnal and cryptic Javan slow loris. We collected 9589 h on six females and 7354 h on six males. Via Generalized Additive Mixed Models, we found that lorises are lunarphobic; they reduce activity levels during cold nights, they have higher activity levels when the relative humidity is close to 100%, and they have high peaks of activity between December and February and between June and August. The activity levels are thus influenced by avoidance of predators, food availability, consumption of insects and nectar, physiological, and behavioral adaptations to cold temperatures and energy requirements during reproductive stages. We highlight the importance of using bio-loggers for cryptic animals as with behavioral observations only, and the observer might underestimate active behaviors and overestimate inactivity.