Rule-based expert systems (RBES) are currently the focus of a great deal of research interest. Most of that work, however, has concentrated on the development of such systems. There has not been much analysis of the resulting RBES. This paper examines two RBESs designed to make audit judgments. The knowledge bases of the initial prototype versions of each system contain the rules used by novice auditors. Each system was refined by having experienced auditors use the system to make the audit judgments for actual clients. The rules contained in the refined versions of each RBES thus represent the knowledge used by an experienced auditor to make a particular audit judgment. The effects of experience are then examined by comparing the rules in the initial prototype knowledge base to those contained in the refined version of each system. Experience appears to provide the capability to deal with exceptions to general rules and expectations.