The diversity, abundance and distribution of reef fish are related to heterogeneity and physical complexity of benthic habitat. However, the field effort required to evaluate these aspects of the benthos in situ, at the scale of entire reefscapes, is greatly constrained by logistical and resource limitations. With moderate ground truthing, both substratum type and seabed topography are amenable to monitoring using satellite data. Here, remote sensing imagery was used to resolve the bathymetry and benthic character of a reef system in Diego Garcia (British Indian Ocean Territory). Replicate fish counts were made at seven measurement stations across the study area using visual census. Monte Carlo simulation revealed that species richness and abundance of several guilds and size groupings of reef fish appraised in situ were correlated with the satellite-derived seabed parameters over areas of seafloor as large as 5,030 m2. The study suggests that satellite remote sensing is capable of predicting habitat complexity at a scale relevant to fish. Furthermore, as larger size classes of fish were better predicted with the satellite habitat complexity data, this technique could be used to predict fish stocks and identify potential sites for marine protected areas where intensive field surveys are not practical.