Preprint of paper published in: TOOLS 2012 - Proceedings of the 50th International Conference, Prague, Czech Republic, May 29-31, 2012; doi:10.1007/978-3-642-30561-0_8 In order to support test suite understanding, we investigate whether we can automatically derive relations between test cases. In particular, we search for trace-based similarities between (high-level) end-to-end tests on the one hand and fine grained unit tests on the other. Our approach uses the shared word count metric to determine similarity. We evaluate our approach in two case studies and show which relations between end-to-end and unit tests are found by our approach, and how this information can be used to support test suite understanding.