In the open ocean, the movements and habitat use of large mobile predators are driven by dynamic interactions between biological and physical variables and complex predator–prey relationships. Understanding the spatial and temporal distributions of pelagic fishes and sharks is a critical component of conservation and fisheries management. Here, we report on a novel non-extractive method for the study of pelagic wildlife, based on baited stereo-camera rigs. The mid-water rigs were derived from existing methodology commonly used in demersal fish surveys. We present new data from 66 moored deployments in Shark Bay, Western Australia (26°10′S, 113°06E) in seabed depths of up to 60 m as a demonstration of the rigs’ ability to resolve spatial variability in pelagic fish and shark assemblages, and to make accurate stereo-measurements of animal lengths. We observed 248 pelagic fishes and sharks from 27 species and 10 families and were able to distinguish between assemblages based nominally on location. We make some general recommendations on optimal deployment protocols and sampling effort regimes, based upon species accumulation rates and times of View the MathML sourceMaxN (maximum number of individuals of a given species in a single video frame). Regression analyses between high quality and low quality stereo-measurements of fish fork-lengths and range were highly significant, indicating that body lengths and distance estimates were consistent even when stereo-measurements were deemed of low quality. Mid-water stereo-video camera rigs represent an efficient tool for the rapid and non-extractive monitoring of pelagic fish and shark populations, with particular relevance for application in no-take marine protected areas.