Relevance feedback has been proven to be a very effective query modification technique that the user, by providing her/his relevance judgments to the Information Retrieval System, can use to retrieve more relevant documents. In this paper we are going to introduce a relevance feedback method for the Bayesian Network Retrieval Model, founded on propagating partial evidences in the underlying Bayesian network. We explain the theoretical frame in which our method is based on and report the results of a detailed set of experiments over the standard test collections Adi, CACM, CISI, Cranfield and Medlars.