We present a formally coherent and consistent multi-species MHC ontology which includes all human MHC alleles and serological groups. The ontology is part of StemNet, a knowledge management system for hematopoietic stem cell transplantation with an integrated semantic search engine. The Owl-encoded MHC ontology contributes to the system in a threefold manner. First, it supports query formulation and query processing as well as mapping onto external terminological resources, second, it eases the interaction with the search engine when navigating through search results, and finally, it provides a formal language for text annotation, a methodological prerequisite for state-of-the-art natural language text processors which are increasingly based on machine learning methods and hence require annotated text corpora.