We investigated the electrolocation performance of the weakly electric fish, Gnathonemus petersii, in novel and familiar environments. By selectively interfering with the fish's sensory input, we determined the sensory channels necessary for navigation and orientation. The fish's task was to locate a circular aperture (diameter: 64 mm) in a wall dividing a 200–1 aquarium into two equal compartments. To assess the fish's performance, we measured (1) the time it took the fish to locate the aperture, (2) the height at which it contacted the divider, (3) its electric organ discharge rate, and (4) the frequency of divider crossings. In the first experiment (novel environment), 50 naive G. petersii assigned to five groups of 10 fish each (intact, blind, electrically “silent,” blind and “silent,” and shamoperated animals) were tested with the aperture presented randomly in one of three positions (aperture center: 7.6, 17.7, 27.8 cm from the bottom). In a novel environment, G. petersii depend on active electrolocation. Despite the changing aperture position, over the 15 trials, fish with a functioning electric organ found the aperture, whereas those without one did not. The electric organ discharge rate was inversely correlated with the amount of time spent searching for the aperture. In a second experiment (familiar environment) 20 intact fish learned the position of a fixed aperture. When we subsequently denervated the electric organ in 10 of these animals, their performance did not differ significantly from that of their conspecifics. Thus, once the fish were familiar with the aperture's position, they no longer depended on active electrolocation. We interpret and discuss this behavior as evidence for a “central expectation” and discuss its possible role in electronavigation.