Information sources in the World Wide Web usually offer two different schemes to their users, an Interface Schema which the user can query and a Result Schema which the user can browse. Often the Interface schema is more restricted than the result schema, moreover many sources offer keyword-search interfaces only. Thus query capabilities of such sources are very small and a useful integration into a mediator-based information system using query capabilities is almost impossible. We propose the Query Tunnelling Architecture for the Wrapping of these restricted web sources. Wrapping of sources by Query Tunneling hides restrictive query interfaces and makes such sources fully queryable based on their result schema. The process of Query Tunneling is divided into two main steps, Query Relaxation to make a higher order query suitable to a restricted interface and Result Restriction in order to filter the results using the original query. Key words: Information Extraction, Information Integr...