This paper presents an agent negotiation model for ontology mapping. We consider that different agents working on the basis of particular approaches arrive to distinct mapping results that must be shared, compared, chosen and agreed. Then, the final mapping result must reflect a better solution than individual approaches.