H ealth care spending accounts for 18 percent of the United States economy and has grown faster than GDP in 42 of the past 50 years. As a result, containing health care costs has become a primary concern of public policy as well as the private sector. Unlike most markets, consumers know very little about prices in health care. Research in other markets has found that increasing price transparency reduces prices (e.g., Brown and Goolsbee 2002; Goldmanis et al. 2010), but it is not clear the same will be true in health care: health insurance insulates patients from prices, reducing the benefits of switching to a lower priced provider.In this paper, I estimate whether access to and use of price information affects the prices paid for medical care. I use a unique dataset in which the employees of a large firm gain access to price information provided by Compass Professional Health Services (Compass hereafter). The novel feature of the data is a direct measure of search: Compass tracks the use of its price information and so directly measures search.I begin by estimating how access to price information affects the prices paid for care. A subset of the employees were given access to price information in a pilot program to determine whether it was worth purchasing for all of the employees. The employees who gained access early did not sign up, volunteer, or select into the pilot program in any way; they were given access early because they worked for the company's corporate offices. I estimate a differences-in-differences model that takes advantage of this variation in access both across groups and over time. The results suggest that access to price information reduces the average price paid by