Mid-water baited remote underwater video systems (BRUVS) are becoming an increasingly popular tool for examining pelagic fish assemblages in a non-destructive, fisheries independent manner. As the technique is relatively novel, critical methodological questions such as the most appropriate attractant for pelagic fish to mid-water RUVS remain unresolved. In this study, we compared the relative effectiveness of 4 attractant treatments (sight: metallic reflectors, sound: bait fish recordings, scent: pilchards and their combination) on the time of first arrival, total abundance of pelagic fish and the relative abundance of 3 pelagic fish species: Trachurus novaezelandiae, Sarda australis and Seriola lalandi. Recordings were made using mid-water RUVS in the Jervis Bay Marine Park, Australia. RUVS using a combination of all attractants recorded the highest abundances and shortest time of first arrival of pelagic fish. This result was primarily driven by Trachurus novaezelandiae. Although not significant, the abundance of Sarda australis was also greatest on the RUVS with all attractants. In contrast, the type of attractant had no effect on the abundance of Seriola lalandi. Bait, the standard attractant used in BRUVS surveys, was a poor performer for pelagic fish in all instances. We suggest that future studies using this sampling method employ multiple attractants.