To evade predators, many prey perform rapid escape movements. The resulting escape trajectory (ET) – measured as the angle of escape direction relative to the predator’s approach path – plays a major role in avoiding predation. Previous geometrical models predict a single ET; however, many animals (fish and other animal taxa) show highly variable ETs with multiple preferred directions. Although such a high ET variability may confer unpredictability, preventing predators from adopting counter-strategies, the reasons why animals prefer specific multiple ETs remain unclear. Here, we constructed a novel geometrical model in which Tdiff (the time difference between the prey entering the safety zone and the predator reaching that entry point) is expected to be maximized. We tested this prediction by analyzing the escape responses of Pagrus major attacked by a dummy predator. At each initial body orientation of the prey relative to the predator, our model predicts a multimodal ET with an optimal ET at the maximum Tdiff (Tdiff,1) and a suboptimal ET at a second local maximum of Tdiff (Tdiff,2). Our experiments show that when Tdiff, 1–Tdiff, 2 is negligible, the prey uses optimal or suboptimal ETs to a similar extent, in line with the idea of unpredictability. The experimentally observed ET distribution is consistent with the model, showing two large peaks at 110–130° and 170–180° away from the predator. Because various animal taxa show multiple preferred ETs similar to those observed here, this behavioral phenotype may result from convergent evolution that combines maximal Tdiff with a high level of unpredictability.Significance StatementAnimals from many taxa escape from suddenly approaching threats, such as ambush predators, by using multiple preferred escape trajectories. However, the reason why these multiple preferred escape trajectories are used is still unknown. By fitting a newly constructed model to the empirical escape response data, we show that the seemingly complex multiple preferred escape trajectories can arise from a simple geometrical rule which maximizes the time difference between when the prey enters the safety zone and when the predator reaches that entry point. Our results open new avenues of investigation for understanding how animals choose their escape trajectories from behavioral and neurosensory perspectives.