Cold-water coral distributions are only partially understood even in the most well-studied areas. This is partly due to the only recent development of appropriate technology, and partly to the high cost and time associated with coral mapping, particularly in deep water. One way to optimise mapping is to develop predictive habitat models as proxies for the actual distribution of corals. These models may provide objective criteria for the selection of prioritised coral mapping areas. In this study, we quantified the relationship between observed cold-water coral distribution and terrain attributes as an important step in developing predictive habitat models. We estimated deep-water coral percentage cover from remotely operated vehicle video and demonstrate how such data can be used to examine quantitative relations between coral cover and terrain parameters (slope, aspect, rugosity and bathymetric position index) derived from ship-borne multibeam swath acoustic data. We show that, at carbonate mound provinces within sites on the Irish margin, coral abundance is correlated with terrain that is strongly sloping and irregular to a varying degree, depending on spatial scale. It is likely that terrain variations influence the hydrodynamic setting, resulting in a varying food supply. A similar approach may be applicable for other fauna in a variety of benthic environments.