A laboratory experiment was performed to see if passively recorded electric signals can be inverted to retrieve the position of fluid leakages along a well during an attempt to hydraulically fracture a porous block in the laboratory. The cubic block was instrumented with 32 nonpolarizing sintered Ag/AgCl electrodes. During the test, several events were detected corresponding to fluid leakoff along the seal of the well. Each event showed a quick burst in the electric field followed by an exponential-type relaxation of the potential distribution over time. The occurrence of these “electric” events was always correlated with a burst in the acoustic emissions and a change in the fluid pressure. These self-potential data were inverted in two steps: (1) using a deterministic least-square algorithm with focusing to retrieve the position of the source current density in the block for a given snapshot in the electric potential distribution and (2) using a genetic algorithm to refine the position of the source current density on a denser grid. The results of the inversion were found to be in excellent agreement with the position of the well where the hydraulic test was performed and with the localization of the acoustic emissions in the vicinity of this well. This experiment indicates that passively recorded electric signals can be used to monitor fluid flow along the well during leakages, and perhaps monitor fluid flow for numerous applications involving hydromechanical disturbances.