Rocky intertidal shores are diverse ecosystems that have been extensively studied, yet there is a surprising lack of knowledge on the methods best suited to quantifying species abundance, diversity and assemblages. We compared visual census, unbaited remote underwater video (mini-RUV), baited remote underwater video (mini-BRUV) and draining/collection methods, with the goal of quantifying the effectiveness of non-invasive methods over more typical invasive methods for quantifying rockpool fish assemblages. In addition, we assessed the optimal set time of video deployment for mini-RUV and mini-BRUV. Fieldwork was undertaken from May–June 2016 using 20 rockpools at seven locations on the SE coast of NSW, Australia. Following standardisation of methods, visual census detected the lowest abundance of fishes whereas hand collection detected the highest out of any method. Increasing sampling time of mini-RUV and mini-BRUV from 10 to 25 min improved their effectiveness, with mini-BRUV detecting comparable abundances to invasive hand collection. The methods did not differ with respect to species richness however. Further, 25 min of mini-BRUV yielded greater detection of common species than hand collection. All together, these results suggest that mini-BRUV is the best method if the goal is to provide accurate estimates of overall abundance, richness and assemblages with minimal disturbance to rockpools, with a preferred set time of 25 min (and at least 15 min). The ability to conduct rapid and non-invasive species surveys of natural rockpool fish communities is an important conservation and management tool given that these habitats can be home to a variety of commercially important and rare species.