The article presents a new method of the optimal design of the electrical rotating machine based on the genetic algorithm which can used for all electrical rotating machine types and to predict their properties. The new method for optimal design allows obtaining the new electrical rotating machine which mass is lower than mass of the conventional electrical rotating machine by two times. As the result of optimal electrical rotating machine design by using the proposed method, the value of the rotor active length is lower by 2.37 times, and the current density is higher by 1.7 times in comparison with the initial electrical rotating machine. The losses are increased by only 25% (power, rotation and materials frequency of both electrical rotating machine are the same). It was also found that the optimality of a particular design scheme, including the rotor magnetic system or the groove type depends on the complex of sizes. Moreover, the change of this complex leads to a change in the optimum design scheme.