Although it was established nearly hundred years ago that fish do hear, it is still not clear how they locate sound sources. In the sixties, directional hearing in fish was considered a close range phenomenon dependent on the lateral line, excluding participation of the inner ear. Twenty years later the view was that teleost hearing specialists possess a vision like, unambiguous three-dimensional directional hearing in the far field, including distance perception. Vectorial analysis of particle motions of the incident sound is a logical prerequisite for directional hearing in fish. This makes neat models for localization of monopole sources, which generates radial particle motions. However, many biologically significant sources are dipoles and multipoles, where nearfield particle motions are not radial, and the generated frequencies are usually rather low. The near field of low frequency sources extends far beyond the range of the lateral line, and may cover most of the range of the inner ear. Updated models for sound source localization should thus also address localization of dipole and multipole sources in the nearfield. A recent model suggests that fish do not instantaneously determine the direction to the source, but are guided to it by keeping constant angle between the body axis and the vectorial sound component. From an historical perspective, it is interesting to note how theories that seemed 'scientifically correct' at the time, may have hampered the development of the field. It is certainly not easy for a terrestrial mammal (man) to imagine the sensory abilities in aquatic species.