The physical environment directly influences the distribution, abundance, physiology and phenology of marine species. Relating species presence to physical ocean characteristics to determine habitat associations is fundamental to the management of marine species. However, direct observation of highly mobile animals in the open ocean, such as tunas and billfish, is challenging and expensive. As a result, detailed data on habitat preferences using electronic tags have only been collected for the large iconic, valuable or endangered species. An alternative is to use commercial fishery catch data matched with historical ocean data to infer habitat associations. Using catch information from an Australian longline fishery and Bayesian hierarchical models, we investigate the influence of environmental variables on the catch distribution of yellowfin tuna (Thunnus albacares). The focus was to understand the relative importance of space, time and ocean conditions on the catch of this pelagic predator. We found that pelagic regions with elevated eddy kinetic energy, a shallow surface mixed layer and relatively high concentrations of chlorophyll a are all associated with high yellowfin tuna catch in the Tasman Sea. The time and space information incorporated in the analysis, while important, were less informative than oceanic variables in explaining catch. An inspection of model prediction errors identified clumping of errors at margins of ocean features, such as eddies and frontal features, which indicate that these models could be improved by including representations of dynamic ocean processes which affect the catch of yellowfin tuna.