In a totally formalized, iterative and incremental software process, each iteration usually implies identifying new requirements, adding them to the current model of the system, checking again the consistency and, in many cases, modifying the model to satisfy all the functional requirements. In this context, the ARIFS (Approximate Retrieval of Incomplete and Formal Specifications) methodology provides a suitable reuse environment (1) to classify, retrieve and adapt formal and incomplete requirements specifications and (2) to reuse the formal verification results linked to them. In this paper, we focus on the first goal describing the classification and retrieval tasks, which are based on functional similarities according to structural and semantic closeness. To this effect, we define four partial ordering relations among reusable components and several measures to quantify functional differences among them. By using these measures, we are able to offer an approximate and efficient retrieval, without formal proofs, and to predict adaptation efforts to satisfy the required functionality.Keywords: Component-based requirements engineering; formal requirements reuse; reuse of verification results; iterative and incremental specification; approximate retrieval. * Partially supported by PGIDT01PX132203PR project (Xunta de Galicia). † Contact author. 607 Int. J. Soft. Eng. Knowl. Eng. 2005.15:607-645. Downloaded from www.worldscientific.com by MONASH UNIVERSITY on 02/03/15. For personal use only. Int. J. Soft. Eng. Knowl. Eng. 2005.15:607-645. Downloaded from www.worldscientific.com by MONASH UNIVERSITY on 02/03/15. For personal use only.