Abstarct: HYSTERESIS motors are self-starting synchronous motors that take advantage of the hysteresis characteristics of the magnetic materials. Robust structure, flat speed-torque characteristics, smooth operation and constant low starting current are the outstanding specifications of these machines. Nevertheless, low efficiency and low power factor are among the disadvantages of common hysteresis motors. Very recently, Coreless Dual Discs Hysteresis Motors (CDDHM) has been introduced to increase power factor and efficiency of the hysteresis machines. Design algorithms of the ordinary machines can be found in many textbooks, but for the new machine due to unavailability of some design or manufacturing experiments, initial design methodology has not been yet proposed. So, employing an optimization algorithm to evaluate dimensions of CDDHM is inevitable. Moreover, conventional optimization approaches are converged normally to a global extremum if the approximated values of the parameters (not far from the optimized values) are available, while is not the case for CDDHM. Briefly, when an initial design is not available, the conventional optimization methods cannot be applied conveniently for the complicated, nonlinear and multi values model of the hysteresis machines. This paper describes a genetic algorithm-baxsed approach with all details, solving successfully the maximum efficiency optimization problem of a CDDHM commencing from any rough initial design parameters. The optimized machine is fabricated and its measured output quantities are compared with those of obtained from the theory. This is given as a validation of the model and design algorithm of the machine.