The seismic performance of reinforced concrete (RC) beam-column joints is typically determined through constant axial loading experiments. The paucity of research in this field is evidenced by the absence of a reliable variable axial force loading protocol, thus resulting in only a small number of experimental studies on the seismic performance of RC joints considering the influence of variable axial force. Furthermore, all such studies assume that the axial force changes in accordance with a simplified linear method, which is an inadequate reflection of the real mechanical characteristics of beam-column joints in RC frame structures under horizontal earthquake excitations. The present study employs finite element simulation of six plane RC frames of seismic intensity 8 (0.2g) with different layers and spans under monotonic and cyclic loading to investigate the change mechanism of joint axial force under horizontal load, and to identify the influencing factors of the change in joint axial force. On the basis of theoretical analysis and statistical regression, a joint variable axial force loading protocol that is consistent with the characteristics of axial force changes and capable of accounting for the nonlinear behavior of materials has been proposed. The results show that the nonlinear behavior of concrete and reinforcement is the primary cause contributing to the nonlinear changes in joint axial force. Additionally, the tensile yield of longitudinal bars at beam end is identified as the principal factor influencing the regularity of the joint axial force changes. The present formula for calculating the change amplitude of joint axial force is proposed, with consideration given to the effects of the following factors: joint type, total number of structural floors, beam span, beam reinforcement area, and beam sectional dimensions. The skeleton curve of the proposed joint variable axial force loading protocol is capable of more accurately reproducing the characteristics of the gradual slowing down of the joint axial force change rate following the tensile yield of longitudinal reinforcement at the beam end of each floor above the joint. Furthermore, the hysteresis rule is able to account for the influence of the nonlinear behaviour of materials on the degradation of unloading stiffness in a reasonable manner.