An adaptive observer is developed for a class of nonlinear systems. Conditions for convergence of state and parameter estimates are presented. The developed theory is used for observer-based parameter identification in the active suspension system of an automobile. A realistic model of the suspension system incorporating the dynamics of the hydraulic actuator is used. The observer is used to adapt on dry friction which is usually present in significant magnitudes in hydraulic actuators. The observer can also be used to adapt on spring stiffnesses, viscous damping and hydraulic bulk modulus. A special adaptive observer is proposed for identification of the sprung mass of the automobile. Since the sprung mass depends on the number of passengers and the load on the automobile, it needs to be regularly updated. The adaptive observers use measurements from two accelerometers and an LVDT. They yield good experimental performance when implemented on a half-car suspension test rig.