Research on corporate bankruptcy is a critically important issue both academically and practically. Although various empirical studies have long been conducted to predict corporate bankruptcy, diverse results have emerged depending on estimation methodologies, selection of explanatory variables, and sample selection issues concerning bankrupt and surviving firms. In this study, we propose a model that can ensure the consistency of estimates by assuming the S_U-normal distribution—a far more flexible distribution than the conventional and highly restrictive normal or logistic distributions—as the error term in a binary dependent variable regression model. Monte Carlo simulation results demonstrated that our proposed S_U-normal distribution model is superior to the conventional logit model in capturing the asymmetry of the error term. Furthermore, when estimating the bankruptcy prediction model for listed companies during the 2000–2005 period, the S_U-normal distribution model proved superior to the logit model in terms of both classification accuracy and prediction accuracy for bankruptcy determination. Substantial differences were also found between the S_U-normal distribution model and the conventional logit model regarding the marginal effects of explanatory variables on bankruptcy probability. These empirical results are significant in that the model proposed in this study enables a more precise quantification of the factors influencing corporate bankruptcy and their ripple effects.