Artificial reefs are often promoted as mitigating human impacts in coastal ecosystems and enhancing fisheries; however, evidence supporting their benefits is equivocal. Such structures must be compared with natural reefs in order to assess their performance, but past comparisons typically examined artificial structures that were too small, or were immature, relative to the natural reefs. We compared coral and fish communities on two large (>400,000 m3) and mature (>25 year) artificial reefs with six natural coral patches. Coral cover was higher on artificial reefs (50%) than in natural habitats (31%), but natural coral patches contained higher species richness (29 vs. 20) and coral diversity (H′ = 2.3 vs. 1.8). Multivariate analyses indicated strong differences between coral communities in natural and artificial habitats. Fish communities were sampled seasonally for 1 year. Multivariate fish communities differed significantly among habitat types in the summer and fall, but converged in the winter and spring. Univariate analysis indicated that species richness and abundance were stable throughout the year on natural coral patches but increased significantly in the summer on artificial reefs compared with the winter and spring, explaining the multivariate changes in community structure. The increased summer abundance on artificial reefs was mainly due to adult immigration. Piscivores were much more abundant in the fall than in the winter or spring on artificial reefs, but had low and stable abundance throughout the year in natural habitats. It is likely that the decreased winter and spring abundance of fish on the artificial reefs resulted from both predation and emigration. These results indicate that large artificial reefs can support diverse and abundant coral and fish communities. However, these communities differ structurally and functionally from those in natural habitats, and they should not be considered as replacements for natural coral and fish communities.