ABSTRACT: Knowledge of the biases and advantages of various methods can inform and create more efficient sampling of either whole fish assemblages or targeted species. Comparisons of stereo baited remote underwater video (stereo-BRUV) and stereo-towed video, a new and relatively uncommon methodology of assessing fish assemblages, were made for assessing fish assemblages beyond diveable depths (>30 m). Stereo-BRUV and towed video footage were analysed for species composition and functional feeding groups in structure-forming macroalgae and sponge-dominated benthic habitats in temperate waters off Warrnambool, Victoria, Australia. Multibeam echosounder (MBES) data, light attenuation and temperature recordings were gathered for each deployment in order to characterise the seafloor structure and oceanographic parameters at each stereo-BRUV location and towed transect. A more abundant and diverse fish assemblage was observed in stereo-BRUV compared to towed video. The fish assemblage observed using the 2 methods was relatively similar in canopy forming algae habitat, but clearly distinct in sponge habitats. Power analysis also showed stereo-BRUV to have greater statistical power when observing total individuals, species richness and functional group richness across both habitats, with the exception of functional group richness in the sponge habitat. Stereo-BRUVs observed higher abundances of elasmobranches, Meuschenia spp., snapper Chrysophrys auratus, demersal invertivores and pelagic planktivores, whilst towed video observed more cryptic and territorial species such as Olisthops cyanomelas and Pempheris multiradiata. Using a combination of stereo baited and towed video sampling techniques records a more detailed description of the fish assemblage on temperate reefs than using either method alone.