There exists a large variety of schedulability analysis tools based on dierent, often incomparable timing models. This variety makes it dicult to choose the best t for analyzing a given real-time system. To help the research community to better evaluate analysis tools and their underlying methods, we are developing a framework which consists of (1) a simple language called RTSpec for specifying real-time systems, (2) a tool chain which translates a system specication in RTSpec into an input for various analysis tools, and (3) a set of benchmarks. Our goal is to enable users and developers of schedulability analysis tools to compare such tools systematically, automatically and rigorously.