This paper reviews the literature on bankruptcy analysis, focusing on the association (the logical linkage) between hazard model and going-concern opinion. Hazard model is a technique that can test the relative impact on the bankruptcy risk of a particular factor, but not a technique that estimates the probabilities of the bankruptcy. Therefore, it is doubtful whether the estimation results by the hazard model can be directly applicable to the going-concern audit. Prior studies repeated the data mining in the absence of theoretical background and rational hypotheses, so the important issues are open to further research, i.e. selection of independent and dependent variables, model selection. Before we apply hazard model to the auditing practice, some hard problems should be resolved. For example, the evaluation method of estimated results and the effects on the firm's behavior by modified opinion are not clarified yet. Though many studies report that the information value of going-concern opinion is suspicious, their research design includes problems in many cases. For the present, no definitive conclusion on the information value exists. From the beginning, it is doubtful whether empirical research using public data can unveil the private decision-making process of auditors, who can know the internal (private) information of the firm.