This paper presents a method for realizing a gear-spring balancer (GSB) that can cope with variable payloads and its application to serial robots. The GSB is constructed with a three-gear train articulating a nonzero-free-length spring to a rotating link with a mass to be statically balanced. In the proposed method, the parameters of the GSB are derived from solving the identity problem of perfect static balancing. The significance of this method is that it enables the GSB to handle variable payloads via energy-free adjustment and allows the selection of spring stiffness. The effectiveness of the proposed method was demonstrated through a numerical example and experimental tests. The analytical results showed that the GSB theoretically achieved perfect static balancing even when the payload varied. The peak actuator torque of the GSB was practically reduced by more than 90% with different payloads. In this work, the application of the GSB to serial robots is described to underline the significance of the payload variability of the GSB in pick-and-place operations. A theoretical model illustrated that the peak actuator torques of a serial robot were reduced by an average of 93.4% during operation.