A central issue in the development of marine research and monitoring programs is the identification of robust method-ologies aimed at detecting spatial and temporal changes in abundance, diversity, and biomass of faunal assemblages (e.g., Maxwell and Jennings 2005; Cheal et al. 2008; Murphy and Jenkins 2010). Decisions on how to assess these metrics are initially based on the type of information required, what spe-cific indices must be measured, the level of precision required to detect change, the repeatability of the method, and the environmental conditions under which assessments will take place (Connell et al. 1998; Willis et al. 2000; Smale et al. 2011). Importantly, the human capital and financial resources avail-able to collect data must also be considered, as costs associated with different data collection techniques can vary substan-tially (Langlois et al. 2010). However, for long-term monitor-ing projects technological advancements, scientific innova-tions, sporadic funding, and changes in the availability of staff expertise often mean that the most suitable assessment meth-ods change through time (Murphy and Jenkins 2010). This may result in time-series datasets that are collected using sev-eral techniques. To ensure that trends in data are not attribut- Abstract Robust assessments of abundance and diversity are essential components of research programs aimed at detecting changes in marine fish assemblages through space and time. This study examined the comparability of Underwater Visual Census (UVC), and Diver Operated stereo-Video (Stereo-DOV) datasets collected across a wide latitudinal range (15°) on coral reefs and temperate rocky shores. There were some differences between methods with regards to both species richness and abundance, with UVC consistently recording higher meas-ures of species richness. Differences were most pronounced at tropical locations where these measures were high. Differences in the characteristics of fish assemblages were primarily driven by UVC differentiating between scarid and pomacentrid species and detecting more cryptic species. When examined at higher taxo-nomic or functional levels however, there was greater comparability between the assemblages recorded by each method, particularly in temperate locations. The UVC method also recorded higher abundances of species tar-geted by fishers in tropical/sub-tropical locations, and subsequently obtained a much higher proportion of length measurements for these species. Data collected using stereo-DOV took 2-3 times longer to obtain than with UVC due to extensive post-processing time required by the stereo-DOV method. This study shows that data collected by the two methods are most comparable in temperate locations, or when examined at higher taxonomic/functional levels. Comparisons should however be more cautious in higher diversity locations, or when assessing at finer taxonomic resolution. When assessing the suitability of either method, availability of time, funding, and relevant expertise should be primary considerations.