ABSTRACT Aim Studies on the geographical distributions of marine species have typically ignored the oceans' inherent 3D nature, instead focusing on two‐dimensional patterns based on surface or near‐surface abiotic conditions. Here, we show the importance of considering oceans as 3D space by comparing fish species richness patterns inferred from 2D and 3D data sources, including species occurrences and environmental data. Location Atlantic Ocean. Time Period Present. Major Taxa Studies Three orders of bony fishes—Beloniformes, Gadiformes and Scombriformes. Methods We generated 2D and 3D species distribution models using two methods: general linear models and simple envelopes. We then summed distributions for each order to map potential species richness in the Atlantic Ocean. Finally, we modelled species richness within each order to examine macroecological patterns based on 2D versus 3D data. Results Our 3D species richness maps provide a different picture of potential distributions of diversity than 2D‐based results, especially for Scombriformes and Gadiformes, which generally inhabit sub‐surface waters. In contrast, 2D and 3D results for Beloniformes, which are generally epipelagic, were quite similar. Using linear models of species richness, we demonstrate disparity among macroecological patterns inferred based on 2D versus 3D data. Main Conclusions Overall, our work emphasises the critical importance of depth in the study of marine ecosystems and the need for detailed depth information associated with organism occurrence records and oceanographic data.