Some bat species seem to be lunar phobic, i.e., they avoid flying in bright areas or during bright periods of the night; however, the evidence is still controversial. We think that part of this controversy comes from pooling data on bat captures and moonlight intensity according to broad categories, such as moon phases, which conceal the high variability among nights. Therefore, we used detailed, long-term field data on three phyllostomid bat species, in order to test the hypothesis of lunar phobia at two different time scales: 1) among nights, by pooling data of different nights according to moon phases and testing for differences in the distribution of captures; and 2) within a night, by analyzing the relationship between capturability and moonlight intensity (measured as illuminance) in one-hour intervals for 29 individual nights. Although most captures of the studied bat species occurred in the first half of the night, their activity pattern varied largely among nights, and was not always unimodal as commonly assumed. At the larger time scale, all studied bat species showed evidence of lunar phobia, as they were more frequently captured on dark moon phases. Nevertheless, at the smaller time scale, only Carollia perspicillata (Linnaeus, 1758) was less frequently captured on brighter periods of the night. We propose that the unimodal activity pattern assumed for frugivorous phyllostomid bats may be an artifact of data organization, and that activity and lunar phobia are much more variable than previously assumed.