Today, software-engineering research and industry alike recognize the need for practical tools to support reverse-engineering activities. Most of the well-known CASE tools support reverse engineering in some way. The Uni-fied Modeling Language (UML) has emerged as the de facto standard for graphically representing the design of object-oriented software systems. However, there does not yet exist a standard scheme for representing the reverse-engineered models of these systems. The various CASE tools usually adopt proprietary extensions to UML and, as a result, it is difficult, or even impossible, to ensure that model semantics remains unambiguous when working with different tools at the same time. In this paper, we examine the capabilities of the two most successful industrial-strength CASE-tools in reverse engineering the static structure of software systems and compare them to the results produced by two academic prototypes. The comparisons are carried out both manually and automatically using a research prototype for manipulating and comparing UML models.