Reefs are being threatened by global warming, natural disasters, and the increased pressure of the global population. These habitats are in urgent need of mapping at high resolution so that these threats can be quantified. Remote sensing can potentially provide such quantitative data. In this article, we attempt to map benthic coral-reef habitats at the Puerto Morelos Reef National Park in Yucatan Peninsula (Mexico) and to assess the accuracy of the technique in providing a baseline data for future monitoring of changes and evolution of the reef system. An IKONOS image was used in combination with checkpoint ground sampling and classified using a supervised maximum likelihood classifier (ENVI 4.5). We show that it is possible to map the reef with acceptable accuracy for the lagoon and discriminate the main habitat types, including vegetation, corals, and bare substrate. But, in areas close to the shore and in the front-reef zone, there were significant misclassifications as well as a failure to delineate spatial structures evident on the ground and in aerial imagery. These difficulties and failures occurred either in the areas deeper than 58 m where depth limits light transmission (particularly in the red channel) or when the spectral response of habitats were too close to be discriminated. This highlights the need to combine these data with other methods, such as acoustic mapping, in order to provide more accurate representations of the benthic habitats of entire reef systems.