Filtering the road unevenness, i.e. comfortable ride for passengers, and vibration and shock isolation for freight, is one of the main thrusts behind the development of active and semi-active suspensions. For control of these systems, a variety of different schemes, mainly from the linear stochastic control area, have been proposed by researchers: and 1/4-car active and semi-active suspension models are used to simulate these schemes. Besides the main input an exogenous input is also considered, namely a velocity disturbance originating from the road unevenness.In most of the literature in this area, this disturbance is considered as white noise, which it hardly is. This a-priori statistical description of the disturbances, is necessary for using stochastic control techniques. In this paper, a comparison of some of these control schemes is performed. The performance and the stability of the controllers are observed, when the assumptions, on which the controller design is based, are not met.More promising trends to control these systems are taking approaches that have been developed in recent years for control of so-called uncertain dynamical systems, i.e. robust control; among them is LQG/LTR. The paper continues with discussing this controller, and introduces the idea of the matching condition on the input disturbances. It turns out that the LQG/LTR control scheme enhances the stability robustness of the system, but at the same time deteriorates the system performance if the matching condition is not met.