Emotional disturbances constitute a major health issue affecting a considerable portion of the population in western countries. In this context, animal models offer a relevant tool to address the underlying biological determinants and to screen novel therapeutic strategies. While rodents have traditionally constituted the species of choice, zebrafish are now becoming a viable alternative. As zebrafish gain momentum in biomedical sciences, considerable efforts are being devoted to developing high-throughput behavioral tests. Here, we present a comparative study of zebrafish behavioral response to fear-evoking stimuli offered via three alternative methodologies. Specifically, in a binary-choice test, we exposed zebrafish to an allopatric predator Astronotus ocellatus, presented in the form of a live subject, a robotic replica, and a computer-animated image. The robot's design and operation were inspired by the morphology and tail-beat motion of its live counterpart, thereby offering a consistent three-dimensional stimulus to focal fish. The computer-animated image was also designed after the live subject to replicate its appearance. We observed that differently from computer-animated images, both the live predator and its robotic replica elicited robust avoidance response in zebrafish. In addition, in response to the robot, zebrafish exhibited increased thrashing behavior, which is considered a valid indicator of fear. Finally, inter-individual response to a robotic stimulus is more consistent than that shown in response to live stimuli and animated images, thereby increasing experimental statistical power. Our study supports the view that robotic stimuli can constitute a promising experimental tool to elicit targeted behavioral responses in zebrafish.