Abstract: The buckling of longitudinal steel bars always accelerates the deterioration of post-peak strength capacity of reinforced concrete (RC) columns. The reliability of the finite element model of RC columns depends on reasonably considering the effect of buckling in the constitutive relationship of bars. The constitutive model of buckled bars proposed by Gomes et al. (G-A model) has explicit physical meaning. However, obvious errors exist in the two basic assumptions. Finite element calculation of cyclic loading tests on the buckling behavior of 24 steel bar specimens was carried out on the OpenSees platform by using fiber section-based plastic hinge nonlinear beam-column elements. Based on the results of the finite element analysis, the errors of the two basic assumptions in the G-A model are analyzed, namely the plastic bending moment M_p and the geometric relationship of ignoring the axial deformation of the member. A modified G-A model is established on the basis of the regression analysis. The results show that the stress distribution on the critical section of a buckled bar is quite different from the plastic bending moment assumption adopted in the G-A model. The errors in calculating the mid-span lateral displacement of buckled bars were introduced by ignoring the axial deformation. The simulation results of the G-A model is less effective. The modified G-A model corrects the plastic bending moment and geometric relationship, and the calculation accuracy isobviously improved.