Marine Protected Areas (MPAs) are a popular conservation strategy aimed at managing anthropogenic pressures and protecting habitats and the diversity of ocean flora and fauna. Robust, cost-effective sampling of fish assemblages is important in understanding the effects of these management strategies on ocean ecosystems. We compared the sampling effectiveness and efficiencies of three commonly used methods of sampling fish assemblages; underwater visual census (UVC), baited remote underwater stereo-video systems (stereo-BRUVs), and diver operated stereo-video (stereo-DOV) and one emerging method (remotely operated vehicle with stereo-video system (stereo-ROV). We assessed the assemblage composition, numbers of species and individuals, and the statistical power to detect hypothetical changes in the number of species and individuals for each method.Stereo-BRUVs sampled a distinctive assemblage compared to all transect-based methods, with more individuals, total species, and predatory fishes from higher trophic groups that are targeted by commercial and recreational fishers. UVC also sampled a distinctive assemblage compared to stereo-ROV and stereo-BRUVs (comparisons with stereo-DOV were not possible due to sampling restrictions). The fish assemblage sampled by UVC consisted of more species and individuals than stereo-ROV and small bodied or cryptic species that were not detected by the video methods. Although stereo-DOV sampled more individuals than stereo-ROV with differences in small schooling species at a few sites, the assemblage composition was broadly comparable. To effectively monitor MPAs a combination of BRUVs, which more effectively sampled fisheries indicator species, and one of the transect based methods should be used. Given the similarities in the assemblages sampled by the stereo-video transect based methods and the advantages associated with health and safety, logistics and field efficiency with remote methods, we recommend stereo-ROV and stereo-BRUVs.