The inverse finite element method (iFEM) [1] is formulated based on the strain-based weighted least-squares variational principle. Its primary advantage lies in utilizing measured strain data rather than predefined natural boundary conditions, thereby enabling structural health monitoring under complex environmental loadings. In this study, a Timoshenko beam-based iFEM model [2] is presented for shape sensing of steel structures under seismic loading. The study begins by identifying an effective strain gauge layout for I-section beam elements and examining the consistency between surface strain measurements and their corresponding section strains. The proposed iFEM model is then adopted to reconstruct structural deformations of steel structures under different static loading conditions, using strain measurements obtained from a limited set of sensors. Subsequently, the seismic record of the 1995 Kobe earthquake is employed as external excitation to evaluate real-time structural responses of a three-story steel frame structure. Based on the numerical comparisons, the proposed iFEM model exhibits strong potential for real-time structural health monitoring.