Danmarks Nationalbank introduced its annual report on financial stability in Denmark in 2000. The purpose of the analyses is to identify risks currently faced by the financial sector. As the stability in the financial sector depends on the customers' financial circumstances, and as the majority of lending from Danish banks is granted to companies in Denmark, analyses of the development in the non-financial sector are crucial in a financial stability context. The primary goal of this paper is to make a tool that can assist the regular analyses of the non-financial sector, namely to make a model that is able to predict the firms that end up in financial distress. As the firms in the non-financial sector may go out of business for various reasons (financial distress, voluntary liquidation, and because they are merged or acquired, etc.) the method of competing-risks models seems appropriate. To get point identification a parametric competing-risks model is suggested. The parametric competing-risks model is estimated and the results are reported. The empirical analysis is based on a panel data set containing information on the whole population of Danish non-financial public limited liability companies (aktieselskaber) and private limited liability companies (anpartsselskaber) that existed between 1995 and 2001, covering around 30,000 firms and more than 150,000 firm-year observations. After application of certain criteria (e.g. exclusion of holding companies and financial institutions), the sample is representative. More than 20 explanatory variables are included in the estimations. Compared to the existing literature this study introduces a number of novel elements to the empirical analysis. Firstly, the empirical distinction between three modes of exit is developed, namely between firms in financial distress, voluntarily liquidated firms, and firms that merge with other firms or are acquired by other firms, and a competing-risks model is estimated. Secondly, a large number of proxies are used for inherently unobservable variables (e.g. uncertainty, ability and motivation). As is discussed in the paper, proxies are important. Thirdly, the extraordinary data set provides an opportunity to compare different specifications of credit risk models, and so the competing-risks specification is compared to a pooled logit model (where all exits are pooled) and to a simple financial distress model (where the exit to financial distress is modelled and all other firms are treated as censored).