A method for determining the orbital parameters of interacting pairs of galaxies is presented and evaluated using artificial data. The method consists of a genetic algorithm which can search efficiently through the very large space of possible orbits. It is found that, in most cases, orbital parameters close to the actual orbital parameters of the pair can be found. The method does not require information about the velocity field of the interacting system, and is able to cope with noisy data. The inner regions of the galaxies, which are difficult to model, can be neglected, and the orbital parameters can be determined using the remaining information.