A novel advanced computational method incorporating the stability functions and the distributed plasticity model into the fiber beam-column element is proposed for the first time to predict the nonlinear inelastic thermo-mechanical behavior of spatial steel frames in fires. The space spread of plasticity is captured by tracing the uniaxial temperature-stress–strain relation of each fiber on sections at integration points. The element stiffness matrix is integrated via the Gauss-Lobatto numerical integration scheme. The nonlinear material behavior is captured by adopting the distributed plasticity model with the force-based interpolation function whereas the geometric nonlinearity of P-δ and P-Δ effects are considered by using the stability functions and a geometric matrix, respectively. The shear deformation and residual stresses which normally exist in steel frames are also considered. For the implementation, a nonlinear thermal incremental-iterative solution scheme based on the Newton-Raphson and Gaussian Elimination algorithm is also developed to address the nonlinear problems due to thermal expansion and material degradation. By taking advantage of the computational benefits of stability functions and the force-based interpolation function, the proposed method offers a novel computational framework for nonlinear structural analysis in a fire with excellent computational efficiency. The reliability and accuracy of the proposed program are verified by comparing the obtained results with results from test data, existing studies, and Abaqus. In addition, the computational performance of the proposed numerical program is about 10 times more effective than the Abaqus commercial software, thus would offer an extremely valuable tool for practical fire resistance designs.