ABSTRACT: Differences amongst diver-based underwater visual census (UVC) methods may influence how fish assemblages are described and interpreted ecologically. Here we compared two common UVC protocols across 32 tropical and 10 temperate reef sites spanning Australia to assess observational biases associated with diver counts. Specifically, we examined how fish data collected whilst deploying a transect tape (reel-out survey) compared to observations made along the transect tape following deployment (pre-laid survey). Each method produced comparable community structures for both biomass and abundance. No differences in species richness, family richness, abundance, or biomass were evident in assessments within realms (i.e., tropical and temperate) or protection status (i.e., fished and no-take). Nevertheless, pooling the data at the continental scale revealed 57% more biomass observed during reel-out surveys than pre-laid surveys, primarily due to more large fishes (≥ 20 cm) from three tropical families and subfamilies: Carcharhinidae, Lethrinidae and Scarinae. Differences in biomass estimates by method altered the contributions of diver-affected families to reef fish trophic structures, with the reel-out method recording more biomass for primary consumers and higher order predators. Overall, our findings indicate tape laying prior to UVC has minimal detectable influence on fish community metrics but highlights that methodological differences can affect the interpretation and reporting of ecological patterns when examined over broad biogeographic scales. Pre-diver baselines of fish abundances, including by application of alternative methodologies such as remote underwater cameras, are needed to disentangle observational biases caused by diver-shy and diver-curious fishes, and thus allow more accurate depiction of reef food webs.