The Stability Graph Method (SGM) is an empirical approach for open stope design, originally developed from extensive case histories in Canada and Australia. It has been applied in Boliden’s Garpenberg and Zinkgruvan mines, providing efficiency in preliminary design and guidance for data collection on key influencing factors. However, uncertainty remains a significant challenge. Geological data are often incomplete at the design stage, and even with detailed geological surveys, heterogeneity in rock mass quality and complex jointing hinder precise parameter quantification. This study develops a probabilistic Stability Graph Method, which integrates probabilistic analysis into the conventional framework. Rather than requiring exact input values, the method uses parameter ranges derived from available geological information. The outputs include probability distributions of stability numbers, providing upper and lower bounds as well as mean values. Combined with empirical stability and caving curves, this approach enables prediction of both stope stability and caving probability. Application to the Garpenberg Mine demonstrates that predicted caving extents agree well with observed performance, validating the method’s effectiveness in supporting risk-informed stope design under uncertainty.