There has been a long discussion about the usefulness of metrics in real-live software projects: Are metrics really helpful and more than a theoretical approach? This talk will present practical experience gained in a special situation: Unknown and poorly documented legacy software had to be maintained by a new team without any contact with the original developers. Instead of a time-consuming review of the existing code, a metrics tool was used to gain an overall impression of the quality and global properties of the code. By this procedure, information about the size, the complexity, the style and the architecture of the system had been gathered with comparatively less efort. Weak code parts could be distinguished from well-engineered ones. On that basis, and as a conclusion, reasonable planning decisions of succeeding maintaining activities had been made.