This paper present the approach and results of a mapping process to define a common metadata format for cross-searching distributed and heterogeneous subject gateways in the EU project Renardus. The outcome in Renardus is a well defined data model with semantic and syntactical definitions of each metadata element. It results in richer and semantically controlled cross-searching. The metadata elements are mainly based on Dublin Core, some further elements and qualifiers are defined in Renardus namespaces. A collection level description schema has also been developed to allow a well structured description of each participating gateway. A Renardus application profile is under development. The Renardus experience and some of its solutions may well be a good basis for similar interoperability efforts.