Visits to the emergency department (ED) are costly, and because some of them are potentially avoidable, some types of ED visits also may be indicative of poor care management, inadequate access to care, or poor choices on the part of beneficiaries. Billings and colleagues developed an algorithm to analyze ED visits and assign probabilities that each visit falls into several categories of appropriateness. The algorithm has been used previously to assess the appropriateness of ED visits at the community or facility level. In this analysis, the authors explain how the Billings algorithm works and how it can be applied to individual physician practices. The authors then present illustrative data from 2 years of Medicare claims data from 5 states. About one third of ED visits are deemed appropriate, and about half could have been treated in a primary care outpatient setting. Another 15% were deemed preventable or avoidable.