In this paper it is shown how a Merton-model approach can be used to develop measures of the probability of failure of individual quoted UK companies. Probability estimates are then constructed for a group of failed companies and their properties as leading indicators of failure assessed. Probability estimates of failure for a control group of surviving companies are also constructed. These are used in probit regressions to evaluate the information content of the Merton-based estimates relative to information available in company accounts and in assessing Type I and Type II errors. Power curves and accuracy ratios are also examined. It is shown that there is much useful information in the Merton-style estimates.