The mining industry has a high fatality rate and a number of non-fatal accidents that make it one of the most hazardous occupations in the world. These accidents often result in deaths, injury, destruction of machinery, and directly impact the productivity of the mines. Therefore, prediction of root cause of these accidents is crucial to take appropriate preventive action plan. In this study, the probabilistic relationship among the accident root causes, effects of accidents, and some additional parameters are developed using Bayesian network (BN) model and construct the conditional probability table (CPT). We consider 224 accidents cases which occurred in Indian underground mine from 2010 to 2023. The three-axiom based sensitivity analysis is used to validate the developed BN model as well as identify the most sensitive parameter that are responsible of the occurrence of accident. Additionally, the proposed research work will help us to develop an intelligence decision support system for safety governance and identify root causes to improve mine monitoring and safety.