Functional demands on quay walls are steadily rising with increases in both ship size and the frequency of port calls, requiring robust safety and performance assessment frameworks. Most existing studies on quay walls remain constrained by simplified analytical or numerical models, limited consideration of failure modes, insufficient model validation, and the absence of site-specific soil correlation, making it difficult to assess real structures under realistic loading conditions. This study presents a novel Cone Penetration Tests (CPT)-driven, performance-based reliability framework that samples directly from measured CPT distributions to preserve field consistent parameter relationships without requiring site-specific covariance matrices. The framework is demonstrated on a recently constructed, full-scale instrumented quay wall in the Port of Rotterdam. A two-dimensional finite element model with the Hardening Soil Small-strain (HSS) formulation is calibrated against
construction-stage monitoring (wall deformations and anchor forces) and coupled to a probabilistic engine to evaluate multiple ultimate and serviceability limit states through an explicit failure tree.

Results indicate that the structural failure of the wall governs the overall reliability of the quay wall, while wall deformation is the most variable response requiring close monitoring. Sensitivity analyses reveal that deeper dredging and higher surcharge loads markedly reduce reliability with wall structural failure governing and serviceability limit states showing the highest sensitivity to these hypothetical changes. The proposed approach provides a generalisable CPT-based methodology for reliability assessment of geotechnical structures based on site investigation data and monitoring data, supporting more informed, data-driven decision-making in design, and life-cycle management.