Blue-green infrastructure (BGI)—engineered, nature-based stormwater controls such as bioretention cells and green roofs—can reduce combined sewer overflows (CSOs). However, hydrological performance can decline over time through sediment deposition and clogging. Despite this, catchment-scale studies commonly assume stationary BGI performance, potentially underestimating CSO volumes and associated risks. This study integrates spatial sensitivity analysis with dynamic deterioration modelling to examine where and when BGI performance loss most affects CSO volumes. A semi-Markov chain model simulated condition-state transitions for individual BGI assets, coupled with EPA SWMM to assess CSO mitigation over time in a realistic case study. Under the modelled parameters, the difference in system-level CSO volume between fully functioning and fully failing BGI reached 18% by the end of the study period. Critically, while most assets exert minimal influence, a subset of high-leverage BGI assets disproportionately governs CSO volumes—primarily determined by asset size and capture ratio. Furthermore, early performance losses are masked by annual rainfall variability, introducing significant delays before failures become detectable through CSO measurements alone. These findings indicate that effective monitoring requires asset-level rather than system-level observation. Targeting a high-leverage subset of BGI can substantially reduce monitoring effort while preserving CSO performance. This study demonstrates how such critical assets can be identified and contributes a methodology for explicitly incorporating BGI deterioration dynamics into catchment-scale models. These findings enable risk-informed monitoring and maintenance prioritisation, treating BGI as a dynamic rather than static component of urban drainage management.