This paper describes the work of a reverse engineering project. The project is concerned with the identification of reusable modules of code from legacy systems. We apply a number of existing techniques and methods to large COBOL programs and attempt to integrate the methods to satisfy the needs of our industrial sponsors. We have found that the issue of module granularity is one of the important factors for successful reuse. By integrating and applying parts of the RECAST method and techniques from the RE2 paradigm, we have identified modules at different levels of granularity. This paper includes some results and a discussion of the identification of reuse modules at different levels of granularity.