The subterranean environment of tunnels poses considerable uncertainty as tunnel structures are ensconced in soil, unlike their above-ground counterparts. This significantly complicates tunnel risk assessment during earthquakes. This study introduces a novel method that integrates multiple damage indices to evaluate the seismic resilience of tunnels. Initially, seismic attenuation is introduced to calculate earthquake exceedance probabilities for various tunnel damage indicators, employing finite element methods (FEM). A robustness evaluation criterion scale value is established based on the amalgamation of multiple tunnel damage indices. Standard Cloud Models are then generated utilising the robustness evaluation criteria. Subsequently, the independent and correlated weights of the robustness evaluation indices are determined using the CRITIC-G1 and decision-making trial and evaluation laboratory (DEMATEL) methods, respectively. A game theory (GT) method is then utilised to amalgamate and allocate weights to these robustness evaluation indices. The evaluation Cloud Models are subsequently generated using a backward cloud generator, based on the division of damage grades for the evaluation criteria and combination weights. Finally, the robustness grade is determined by comparing the similarities between the standard and evaluation Cloud Models. The repair time of the tunnel is quantified using a repair function based on robustness grades. The efficacy of the seismic resilience assessment method is discussed based on three hypothetical cases, providing valuable guidance for assessing the seismic resilience of underground structures.