There are over 100 million visits to emergency departments in the United States annually that could be a source of data for multiple uses including disease surveillance, health services research, quality assurance activates, and research. The patients' motivations for seeking care or the reason for visit (RFV) are recorded in every case. Efforts to utilize this rich source of data are hampered by inconsistent data entry and coding. This study analyzes ICD-9-CM, SNOMED-RT, and SNOMED-CT encoding of the RFV for accuracy. Each encoded reason for visit was compared to the text entry recorded at the time of visit to determine the closeness of fit. Each coded entry was judged to be an exact lexical match, a synonym, a broader or narrower concept or no match. SNOMED-CT was a lexical match or synonym for 93% of the text entries, while SNOMED-RT matched 87%, and ICD-9-CM matched 40%. We demonstrate that SNOMED coding of the RFV is more accurate than ICD-9-CM coding.