We investigate the applicability of classical resolution-based theorem proving methods for the Semantic Web. We consider several well-known search strategies, propose a general schema for applying resolution provers and propose a new search strategy ”chain resolution” tailored for large ontologies. Chain resolution is an extension of the standard resolution algorithm. The main idea of the extension is to treat binary clauses of the general form A(x) ∨ B(x) with a special chain resolution mechanism, which is different from standard resolution used otherwise. Chain resolution significantly reduces the size of the search space for problems containing a large number of simple implications, typically arising from taxonomies. Finally we present a compilation-based schema for practical application of resolution-based methods as inference engines for Semantic Web queries.