Augmented Reality for Interventional Guidance has first been proposed more than 10 years ago. The early systems demonstrated the principle. The technology available at the time, though, did not support the design of practical systems. This has changed - computer performance has increased dramatically, and practical AR systems are indeed possible. It has also become easier to assemble a functional AR system, and the community that investigates medical AR is growing. We outline specific challenges of AR systems in medicine, and we discuss strengths and limitations of basic design options, reviewing some of the existing medical AR systems. At present, there are no products in the field of medical AR. We argue that the focus of the research needs to be very much application driven.