In recent industry, self-driving vehicles are very actively researched because they are considered as a new industrial power in most countries. In self-driving vehicles, many different fields should be researched at the same time. The research of road database for self-driving vehicles should not be indispensable in those fields. In order to design the road database for self-driving vehicles, the conventional road database should be enhanced in many sides, because it is designed as auxiliary equipment under assumption that a human is driving the vehicle. While the conventional road database are focused on the accumulation of static data, the road database for self-driving vehicles must include dynamic data such as roads that may be temporarily closed because of repairing. This paper proposes the design of a road database for self-driving vehicle so that it includes dynamic data as well as static data on roads. In order to provide a generality and scalability of road database, we design the database by using the well-known Entity-Relationship Model. In order to simplify the database, we also use the abstracting method and then extracted 6 entities and 10 relationships. From these entities and relationship, the entity-relationship diagram is also proposed.