Abstract: To investigate the influence of longitudinal force loss on the longitudinal seismic performance of shield tunnels with internal structures, a three-dimensional finite element model of shield tunnel with and without internal structures was established. The model was based on an extra-large cross-section double-deck shield tunnel for combined highway and railway use, passing through a soil-rock composite stratum. The implicit dynamic time-history method was employed to analyze the opening deformation of the ring joint, structural damage and internal force distribution of the shield tunnel under longitudinal seismic excitation. The analysis considered varying degrees of longitudinal force loss and the influence of internal structures. The results indicate that: (1) Longitudinal force loss reduces the longitudinal stiffness of the tunnel, leading to increased ring-joint opening deformation under longitudinal earthquakes. Specifically, the crown opening increased by 1.2 mm at the soil-rock interface and by 2.49 mm at the deformation joint, while the invert opening increased by 1.91 mm at the deformation joint. Internal structures enhance the longitudinal stiffness of the tunnel and reduce ring-joint openings, but they increase the ring-joint opening deformation at the deformation joints. (2) The extent of structural damage in the tunnel increases with longitudinal force loss. Damage to internal structures primarily occurs in the cast-in-place sections and tends to propagate along the segment ring joints. (3) The maximum positive bending moment of the tunnel occurs at the soil-rock interface, while the maximum negative bending moment occurs at the deformation joints. Negative axial forces appear within a range of 15 rings on the soft soil side of the soil-rock interface. (4) Both longitudinal force loss and the inclusion of internal structures reduce the peak internal forces in the tunnel under seismic action. When longitudinal force is not lost, the internal structures reduce the maximum positive bending moment by 19.22%, the maximum negative bending moment by 10.59%, the maximum positive axial force by 8.39%, and the maximum negative axial force by 6.15%.