Over the past 4 decades, Hawaii has developed a system of Marine Life Conservation Districts (MLCDs) to conserve and replenish marine resources. These MLCDs vary in size, habitat quality, and management regimes, providing an excellent opportunity to test hypotheses concerning marine protected area (MPA) design and function using multiple discreet sampling units. Digital benthic habitat maps for all MLCDs and adjacent habitats were used to evaluate the efficacy of existing MLCDs and adjacent habitats using a spatially explicit stratified random sampling design. Most fish assemblage characteristics (e.g. species richness, biomass, number of individuals) were highest in colonized hard bottom habitats (>10% live coral cover), followed by uncolonized hard bottom habitats (<10% live coral cover), macroalgae, and sand, respectively. Although biomass was low in sand habitats, apex predators accounted for 62% of the biomass on sand within MLCDs, highlighting the importance of this habitat in reserve design. Within habitats, values for assemblage characteristics were typically higher for MLCDs compared with adjacent areas, emphasizing the importance of protection from fishing irrespective of habitat. Rugosity explained much of the variability in species richness and biomass across all locations. Overall, MLCDs protected from fishing, with high habitat complexity and good habitat quality (e.g. high coral cover and low macroalgae cover), had higher values for most fish assemblage characteristics. Integrating mapping and assessment of reef fish habitat utilization patterns allowed for a robust approach to MPA evaluation and can help inform decisions about MPA design and effectiveness, as well as helping to define essential fish habitat and ecosystem function.