It is possible to present an unstable driving state to a driver even when learning of normal driving characteristics is not completed. A first driving instability determining unit of an information providing unit (100A) estimates driving instability based on a difference value between plural traveling state distributions of different time ranges on the basis of acquired traveling state data. A second driving instability determining unit of the information providing unit (100A) estimates driving instability through the use of a process different from the estimation process used in the first driving instability determining unit. A learning completion determining unit of the information providing unit (100A) determines that the learning is completed when a predetermined learning time elapses from the start of collection of the traveling state data, depending on a degree of learning (SD) at which the traveling state distribution calculated by a first traveling state distribution calculating unit is matched with the driving characteristic of a driver. An instability selecting unit of the information providing unit (100A) selects the instability estimated by the first driving instability determining unit when the learning is completed and selects the instability estimated by the second driving instability determining unit when the learning is not completed.
