This paper aims to develop a new approach to case-based reasoning without similarity constraint. The key to this is the case relation model which enables identification of relevant cases from a global perspective. Fuzzy linguistic rules are adopted as powerful means to represent knowledge about relevance between cases in the case relation model. The construction of fuzzy relevance rules can be realized by learning from pairs of cases in the case library. The empirical studies have demonstrated that our CBR system using fuzzy relation model can work with an extremely small number of cases while still yielding competent performance.