Multispecies predator aggregations are among the most spectacular events in the open ocean. These transient aggregations play a critical role in pelagic food webs and support globally important fisheries and ecotourism industries. However, despite their biological and economic importance, little is known about the behavioural mechanisms by which these aggregations are formed, persist or ultimately break down. Thanks to advances in filming techniques, computer vision and automated tracking, these predator aggregations can now be transformed into quantitative behavioural datasets. Here, using fine-scale tracking of striped marlin (Kajikia audax), sailfish (Istiophorus platypterus) and mahi mahi (Coryphaena hippurus), I quantify how individual trajectories and neighbour interactions shape multispecies predator aggregations. Tracking revealed species-specific attack strategies, spatial distribution around prey schools and rapid colour change, while analyses of inter-predator interactions showed that the identity, abundance and proximity of neighbouring predators not only influences prey access, but in some cases hunting success of the whole aggregation. Ultimately understanding these fine-scale behavioural mechanisms is fundamental to predicting how pelagic predator communities may respond to changing ocean conditions and thus has important implications for fisheries management and the conservation of pelagic ecosystems.