Numerous sampling methods have been applied to study intertidal rock pool fishes. There is limited information comparing the performance and suitability of such methods, which complicates comparisons between studies. We compared the performance of the bailing and visual census methods in 10 rock pools to quantify abundance, species richness, and structure of fish assemblages in rock pool intertidal systems considering fish residency affinities, rock pool volume, and substratum rugosity. Sampling was conducted 13 times across the complete set of 10 rock pools from June to December, 2018. Each of the pools was sampled with a visual census and then the bailing method. A sampling event was defined as the process of conducting both a visual and a bailing survey in a single rock pool. In total, 1,749 individuals of 14 species were detected using the bailing method, whereas 438 individuals of 8 species were recorded via visual census. The bailing method yielded higher abundance and species richness of resident, opportunist, and transient fish than the visual census. The detection of individuals and species showed a positive association with rock pool volume and negative association with substratum rugosity for both methods. In 32 sampling events, the bailing method detected fish correcting reports of fish absence (false negatives) recorded with the visual census. For the bailing method, resident fishes dominated the fish assemblage, whereas, for the visual census, resident and opportunist fishes contributed similarly to the community.