This study conducted the first high-resolution investigation of the ichthyofaunal assemblages on a high-latitude coral reef in the Western Indian Ocean (WIO). Two-Mile reef, in South Africa, is a large, accessible patch-reef, and was selected as a candidate study area. Although the effect of season in structuring coral reef fish communities is most-often overlooked, the relationship between these fish communities and their habitat structure has been investigated. In South Africa, however, neither of these potential community-level drivers has been explored. As coral reefs worldwide are faced with high levels of usage pressure, nondestructive underwater visual census (UVC) techniques were identified as the most appropriate survey methods. This study had two primary aims that were; (1) to identify the most suitable technique for the UVC of coral reef fishes, and to test variations of the selected technique for appropriateness to implementation in long-term monitoring programs, and (2) to determine if possible changes to ichthyofaunal community structure could be related to trends in season and/or habitat characteristics. A review of the literature indicated that the most appropriate UVC method for surveying epibenthic coral reef fishes is underwater transecting. To compare the traditional slate-based transects to variations that implement digital image technology, slate transects were compared to a first-attempt digital photographic transect technique, and digital videographic transects. Videographic transects produced the most favourable species richness, abundance, and standard deviations of the three techniques. Diversity was not significantly different between transect techniques. The minimum required sample size was lowest for videographic transects (17 replicates), intermediate for photographic transects (27 replicates) and highest for slate transects (37 replicates). Videographic and photographic transects required greater analysis time to generate counts, but required lower observer training time. While videographic transects produced the lowest proportion of species considered unidentifiable, all three transect techniques showed similar functionality to surveying epibenthic coral reef fishes. Videographic transects were therefore identified as the most appropriate UVC technique for this study. Videographic transects at shallow (6 – 14 m), intermediate (14 – 22 m) and deep (22 – 30 m) depths in mid-winter and mid-summer, sampled a total of 41 families consisting of 209 species and 18172 individuals, dominated by pomacentrids in abundance and labrids in richness. The fish assemblages on Two-Mile Reef were found to be similar in composition to lower-latitude WIO reefs. Overall ichthyofaunal abundance and richness was significantly higher in summer than in winter, and was higher at shallow sites than at intermediate and deep sites. A multivariate approach confirmed differences between seasons at shallow depths but not between seasons at intermediate and deep depths. The fish assemblages on Two-Mile Reef can therefore be described as being comprised of four relatively distinct communities: a shallow, winter community; a shallow, summer community; a year-round intermediate community; and a year-round deep community. The distributions of discriminating species indicated that high abundances of the algal-feeding pomacentrids are observed only at shallow and intermediate sites while high abundances of the zooplanktivorous serranid subfamily, the Anthiinae, are observed predominantly at deep sites. Assessment of all measured supplementary variables indicated that of all factor combinations, observed patterns could be ascribed most strongly to depth. Quantification of reef characteristics indicated that as depth increases, habitat complexity decreases, benthic communities shift from dense coral domination to sparse sponge domination, and algal biomass and cover decreases. The ability of the videographic transect technique to detect changes in community structure with season and depth indicates that season and depth should be accounted for in future high-latitude ichtyofaunal surveys, and that the videographic transect technique is suitable for implementation in long-term monitoring programs on coral reefs. The similarity in fish assemblages between Two-Mile Reef and lower latitude regions suggests that the protocol for surveying epibenthic coral reef fishes, resulting from this study, is relevant throughout the continental WIO.