We describe the information technological challenges in the project ArcheoInf which aims at providing a unified access to heterogeneous archaeological databases. We consider the difficulties caused by legacy databases not designed for structured data retrieval and propose an ontology-based solution to this problem. In particular we present a process that migrates the data into an ontology-based triple store system, which can be queried with SPARQL. The problem of heterogeneous vocabularies is solved by a mapping of the respective datafields into an archaeology-specific thesaurus which is also kept in the triple store. Our approach is expanded by the possibility to integrate further SQL-databases into the ArcheoInf-system without migrating them into the triple store, provided they are based on a datamodel which is suitably compatible to the ontology we use.