This study investigates the ability of publicly available failure‐prediction models to predict the failure of companies listed on the Australian Stock Exchange. The failure‐prediction models identified in the 1990 version ofAUP 17 were evaluated and re‐estimated using recent Australian data. In addition, a model was developed incorporating publicly available information not previously included in Australian failure‐prediction models. Variables measuring the size of the board of directors, changes to the board, the lag in reporting accounts, and average effective tax rates were added to the variable set. Most of the earlier models, the re‐estimated models and the expanded model significantly outperformed chance, which suggests that auditors should consider using such models as part of their going‐concern assessment