Settlement-stage larvae of five pomacentrid species were tested for the ability to hear and localize reef sounds in a nocturnal field experiment at Lizard Island, Great Barrier Reef. Linear selection chambers allowed larvae to swim either toward or away from nocturnal reef sounds from an underwater speaker 25 m away. Data were analyzed under two assumptions: 1) each individual larva in a selection chamber acted independently, and 2) each individual larva did not act independently. Two-thirds of Chromis atripectoralis Welander and Schultz, 1951 larvae swam toward the reef sound, and this tendency was significant regardless of the assumption of independence. Larvae of C. atripectoralis were also more likely to swim toward the speaker when it was broadcasting sound than when it was quiet. Neopomacentrus cyanomos (Bleeker, 1856) larvae also swam toward the sound, but this was statistically significant only under assumption 1. Three Pomacentrus species did not swim toward the sound, although under assumption 1, one was significantly more likely to swim away from it. These results seem to show among-species differences in the response to sound, but the possibility of temporal differences in behavior cannot be ruled out. Larvae of at least some pomacentrid species can use reef sounds to localize a sound source at night, providing evidence that sound emanating from reefs at night is a useable sensory cue for fish larvae trying to find settlement habitat.