This paper reports on MIMOR which was modelled as an open meta infor-mation retrieval system at the handling of diverse information retrieval ob-jects. It is used to gain further insights concerning the behaviour of informa-tion retrieval systems as well as for teaching in the field of Information Sci-ence. The main issue of MIMOR is the exploitation of users’ relevance feed-back in order to optimise the fusion of several retrieval engines or resources. As yet, MIMOR has been applied to a domain specific collection, to a multi-lingual corpus, and to music data. Experiments within the CLEF evaluation framework and additional ones are discussed here. Key words: information retrieval, fusion, relevance feedback, evaluation, meta system, multilingual information retrieval