The importance of predicting bankruptcy risk of firms is increasing because of later financial crisis. Despite practical researchers trying to present models for predicting this risk, it seems that an optimum and acceptable model that is reliable for financial statement users and auditors in order to increase their ability in decision making and professional judgment has not been presented yet. Therefore more researches in this region can help to understand financial crisis and bankruptcy risk of firms and finally increase the probability of accessing such a model. This research investigates 31 financial ratios have been used with Bayesian approach to test market information of firms listed in TSE during 1386 to 1396. Finally a model consist of 6 variables (retained earnings to total asset, changes in net income to the sum of absolute value of two years net income, financial leverage, equity book value to debt book value, excess of debt to total asset and loss index) has been designed and presented for predicting this risk. The ability and prediction accuracy of this model has been tested by Receiver Operating Characteristics (ROC) curve and actual information of crisis and non-crisis firms. The results show high precision and accuracy of the model.