This paper addresses the design of traffic state estimator for traffic flow in motorway stretches. Motorway traffic dynamics is modelled by a stochastic macroscopic traffic flow model, while a traffic state estimator is designed based on the extended-Kalman-filtering methodology. Besides the estimation of normal traffic variables, the developed traffic state estimator deals with the on-line estimation of unknown model parameters. Simulation investigations were conducted to test the traffic state estimator under various traffic conditions in a motorway stretch.