A novel relational routing scheme is proposed to solve the problem of querying and aggregatingsemantic media content on the network at an effective pace. The scheme can route the semantic media query andquickly backtrack relevant content from the network based on the named media and semantic association. First, themodels of semantic media and relational query are introduced, along with the necessary data structures andalgorithms for relational routing protocol. In particular, the relational matching algorithm, the procedure ofrelational routing and the approach for avoiding incomplete response of pending relational query are elaborated on.Second, the key problems of relational routing are discusse, such as media naming, query preference and theapplication. Finally, the experimentation of relational routing model and algorithms on real platform are made.Results show that the scheme is highly efficient for retrieving semantically relevant content and provides aneffective approach for semantic aggregation of distributed and dynamic media.