This paper presents a method based on genetic algorithm for the automatic generation control of power systems. The technique is applied to control a system, which includes two areas tied together through a power line. As a consequence of continuously load variation, the frequency of the power system changes with time. In conventional studies, frequency transients are minimized by using integral controllers and thus zero steady-state error is obtained. In this paper, integral controller gains and frequency bias factors are determined by using genetic algorithm. The results of simulation reveal the application of the genetic algorithm having easy implementation to find the global optimum values of the control parameters.