Abstract: Background and Objective: Roof fall is one of the greatest single hazards faced by underground coal miners. This accident may have detrimental effects on studyers in the form of fatal and non-fatal injuries as well as downtimes, equipment breakdowns, etc. Due to different impacts of contributing parameters on roof fall and ill-defined or even immeasurable nature of such factors, this problem is an uncertain and complex issue. As a result, development of a methodology for roof fall risk evaluation under uncertainty condition has a remarkable role on safety of underground coal miners. Methodology: This study proposes a new quantitative assessment framework, integrating the inference process of Bayesian networks and fuzzy set theory with the traditional probabilistic risk analysis. The constructed Fuzzy Bayesian Network (FBN) based model has 12 root nodes contributing to the failure of the leaf node. The geology maps and data related to mining equipment at Tabas Coal Mine (TCM) are used to determine the prior probability of FBN root nodes. In addition, weighted sum algorithm is used to populate the conditional probability table of intermediate and leaf nodes. Results: The new model quantifies uncertainty in roof fall and also provides an appropriate method for modeling complex relationships in underground mining. Conclusion: Finally, the proposed approach is illustrated with an application for the TCM and found to be a powerful technique for coping with uncertainties and predicting roof fall risk.