Evaluating line criticality in public transport networks is computationally challenging due to overlapping routes, frequency-based operations, and the combinatorial growth of multi-line disruption scenarios. Traditional approaches, whether topology-based or reliant on repeated equilibrium assignments, do not scale to realistic networks, where full-scan or multi-failure enumeration can require thousands of computationally intensive assignments. This study presents an innovative, computationally efficient framework that integrates a section-based transit assignment model with Mutual Information–based Global Sensitivity Analysis (MISA) to rank individual line criticality without simulating any failure scenarios. Stochastic O–D demand samples are generated using Monte Carlo methods, and equilibrium flows are computed once per scenario using a variational inequality formulation on an augmented network representation. The resulting dataset is analyzed using information-theoretic sensitivity indices, capturing nonlinear and interaction effects while reducing the computational burden by several orders of magnitude compared to brute-force multi-line failure analysis. Application to a benchmark network demonstrates that the proposed method reliably identifies lines whose operational roles and network embeddings make them most influential for system-wide performance. The framework offers a practical, scalable solution for infrastructure planners seeking data-driven tools for resilient public transport design and prioritization under uncertainty.