A major task that auditors undertake in formulating their audit opinion is to estimate the dollar error that might exist in accounts in light of their evaluation of internal control strengths and weaknesses. The research that has been undertaken to evaluate the quality of auditor performance on this task, however, has produced mixed results. The current study tries to overcome some theoretical limitations and measurement limitations that have undermined prior studies. Forty practising auditors rated their abilities on fourteen characteristics derived from a model of expertise. They then undertook an experiment where they judged the extent of dollar error that might exist in the inventories of a manufacturing company. They based their judgments on working papers provided to them that contained, among other information, an evaluation of the company's internal control system. An estimate of the ‘true’ dollar error in the inventories was calculated using a program that simulated the company's internal control and accounting system. This estimate was then used to calculate the accuracy of the auditors’ judgments. Their ratings on the expertise characteristics were not related to their judgment accuracy. Their ratings were related, however, to their confidence in their judgment accuracy. Auditors who considered themselves more expert at the task were more confident in their judgment accuracy but they were, in fact, no more accurate than auditors who considered themselves to be less expert at the task.