Abstract
Travel times for links in urban networks are governed by the stochastic process because of random traffic congestion. Therefore, a traffic model cannot compute accurately travel times on links in terms of deterministic value. In order to calculate more realistic travel times, a traffic model should incorporate random travel time variations into the deterministic one. In this paper the authors introduce the concept of reliability for reflecting travel time variations and driver's travel time uncertainty and build the reliability assignment model. Reliability assignment model can reflect driver's behavior, try to minimize travel time variations and their own uncertainty in terms of the perceived travel time.