Stochastic preview control with time-delay consideration was investigated for an active vehicle suspension system with look-ahead sensor. The road roughness height was modeled as a filtered white noise stochastic process and the rate of change in the road profile at some distances in front of the vehicle was measured by the look-ahead sensor. The suspension system was optimized by minimizing the performance index containing the mean-square values of body acceleration, suspension rattle space, tire deflection and control force. Only partial state variables were assumed able to be measured and the measurement noises were taken into consideration. Design of the preview compensator that may be called stochastic optimal, output feedback, preview regulator problem with time-delay consideration was reduced to the classical linear quadratic Gaussian (LQG) control problem, by converting the state equations and the performance index into discrete forms and augmenting the state and output vectors. Simulation results demonstrated that time delay should be considered in the stochastic preview control of an active suspension system, especially when large time delay exited. If the suspension system with time delay was controlled by the optimal controller designed with no consideration of time delay, instability in responses might occur and the control effects possibly got worse with preview time increase.