Abstract: To address the dual challenges of empirical dependency in fuzzy parameters and insufficient spatial characterization of failure mechanisms in fuzzy measurement theory, a stability analysis method for open-pit mine slopes is proposed by integrating BP neural networks and fuzzy measurement theory in this study. A three-layer BP neural network-based proxy model for fuzzy parameters is developed, establishing a BP-Fuzzy slope instability probability assessment model with self-learning capability. Taking the Gao Village Phase II open-pit iron mine slope as a case study, the results demonstrate that the proposed model predicts a spatially decreasing trend in slope instability probability, and reveals localized rockfall risks consistent with field-observed crack. Furthermore, finite element strength reduction method was employed to precisely identify failure locations and elucidate damage mechanisms, while the traditional Bishop method is used to provide cross-validation. A "probability assessment–slip surface localization" comprehensive evaluation system is thereby constructed, which provides a quantifiable and visualized solution for dynamic risk assessment of open-pit mine slopes.