Natural lightning strikes induce impulsive overvoltages on transmission lines and its terminal equipment. These overvoltages may cause failures in insulation mechanisms of electrical devices in the power system. It is important to test the insulation strength of a device against these impulsive overvoltages. Usually, Marx generators are used to generate impulse waveforms for testing purposes. A novel approach is proposed to obtain resistor settings of a Marx generator for impulse testing of power transformers. This approach enables us to overcome most of the major challenges in the commonly used trial-and-error method, including excessive time consumption and potential damage to the transformer. The proposed approach uses the frequency response of the transformer to synthesize a circuit model. Then, a genetic algorithm based optimization-enabled electromagnetic transient simulation approach is used to obtain the resistor settings. The proposed approach is validated by a real impulse test conducted on a three phase power transformer.