Keeping up with changes in source system terms in a local health information infrastructure requires substantial effort. I developed a program to assist us that returns candidate mappings based on string similarities between newly encountered source test names, existing source test names, and our master dictionary term names. I evaluated this program's performance in identifying correct mappings through a retrospective study of term mappings to our master dictionary from four radiology systems. For source terms created after the initial system integration, the semi-automated mapping program identified correct mappings for 76.3% of terms from all systems. Overall, the program correctly identified mappings for 45.6% of all terms by exact string match to an existing term. The program identified correct mappings for 36.9% of the terms without an exact string match by string comparison to existing source terms, and for 54.4% of the remaining unmapped terms by string comparison directly to master dictionary terms. Because managing vocabulary mappings is resource-intensive, accurate automated tools can help reduce the effort required for ongoing health information exchange among disparate systems.