One task of coral reef remote sensing is to obtain the composition and distributionof benthic categories. However, there is still a great deal of uncertainty anddifficulty to discriminate reef benthos by means of remote sensing owing to thespatial heterogeneity and complicated spectrum of coral reef. Spectral characteristics of different coral reef benthos are the basic prior knowledge for remote sensing of coral reefs. Based on in situ spectral data and simulated satellite data, this paper analyzed the spectral characteristics of different coral reef benthos, especially the spectral properties of different coral types. The influence of coral pigments on coral spectra was also analysed. Finally, four kinds of commonly used satellite data(Landsat 8, IKONOS, Quickbird and SPOT 5) were simulated to investigate the spectral separability of different reef benthos from space. Results showed that sand and bleached corals could be easily identified by the reflectance curves in the visible bands. The first-order spectral derivation in visible bands was a good way to distinguish algae, seagrass and healthy corals. The differences of families, genera, species, coral shapes and coral colors would have obvious impact on the spectral characters of corals. In addition, Chlorophyll contents (including Chlorophyll-a, Chlorophyll-b and Chlorophyll-c) had high correlativity with reflectance of corals, which would exert notable influence on coral spectral features. Zooxanthella had the similar influence, but not as obviously as that of chlorophyll. Its density would affect the peak features of coral reflectance. Among the commonly used multi-spectral satellite data, Landsat 8 had the ability to distinguish sand, bleached corals, algae, healthy corals and seagrass owing to its coastal band, while IKONOS and Quickbird could identify sand, bleached corals and seagrass. Comparatively, SPOT5 had a poor performance, which could only identify sand and bleached corals. However, in the identification of different types of corals, multi-spectral satellite data failed to capture the elaborated spectral features and hyperspectral data with high spatial resolution was needed for effective identification. In the future work, we will further expand more coral reef benthos samples and establish the spectral database of coral reef to provide the data support for the establishment of coral reef monitoring system in China.