Marine environments at depths below −10 to −25 m are almost devoid of ambient red sunlight because water quickly attenuates long wavelengths. This stenospectral light environment presents unique opportunities for organisms that can transform ambient blue-green light into red light by fluorescence. Numerous marine fish species display intricate patterns of fluorescence. Because color vision is a key component of fish sensory ecology, several putative visual functions of red fluorescence have been proposed but are difficult to test experimentally. Here, we follow a comparative approach to assess the consistency between the phylogenetic distribution of red fluorescence with its presumed functions. We collected and analyzed the largest data set of red fluorescence in fishes to date, consisting of confirmed cases in 272 primarily diurnal fish species from 49 out of 90 surveyed fish families and 12 out of 21 surveyed fish orders, contrasted to 393 fish species with confirmed absence of red fluorescence. Based on a priori hypotheses on adaptive function, we compare the prevalence of red fluorescence among pre-defined sets of species based on ecological or biological characteristics while controlling for shared ancestry. When comparing between species, we find no evidence that red fluorescence is more prevalent in deep-water species, contrasting with our recent finding that fluorescence brightness increases with depth within species. There is also no evidence for a role in group-driven communication. Phylogenetic patterns are consistent, however, with three other predictions. First, fluorescence with a rather patchy distribution across the body occurred significantly more often among sit-and-wait predators or otherwise sedentary fish than in more mobile species, consistent with background matching for camouflage. Second, small, predatory fishes tended to show red fluorescent irides disproportionally often consistent with a proposed function in prey localization. Finally, sexually dimorphic species showed fluorescent fins more often, as predicted if relevant in sexual communication. From these findings, we derive predictions for experimental investigations of the presumed functions of red fluorescence.