Coral reefs are biodiversity hotpots that are under significant threat due to the degradation and death of hard corals. When obligate coral-dwelling species die, the remaining species must either move or adjust to the altered conditions. Our goal was to investigate the effect of coral degradation on the ability of coral reef fishes to assess their risk of predation using alarm cues from injured conspecifics. Here, we tested the ability of six closely related species of juvenile damselfish (Pomacentridae) to respond to risk cues in both live coral or dead-degraded coral environments. Of those six species, two are exclusively associated with live coral habitats, two are found mostly on dead-degraded coral rubble, while the last two are found in both habitat types. We found that the two live coral associates failed to respond appropriately to the cues in water from degraded habitats. In contrast, the cue response of the two rubble associates was unaffected in the same degraded habitat. Interestingly, we observed a mixed response from the species found in both habitat types, with one species displaying an appropriate cue response while the other did not. Our second experiment suggested that the lack of responses stemmed from deactivation of the alarm cues, rather than the inability of the species to smell. Habitat preference (live coral versus dead coral associates) and phylogeny are good candidates for future work aimed at predicting which species are affected by coral degradation. Our results point towards a surprising level of variation in the ability of congeneric species to fare in altered habitats and hence underscores the difficulty of predicting community change in degraded habitats.