The application of optimal anticipatory control assumes that the entire state is available for feedback. However, in the case of injection molding this assumption is violated due to economic considerations. For that purpose, this paper discusses the effectiveness of optimal anticipatory control using estimates of the states obtained from a dynamic system, called observer. State estimation is essential to optimal control techniques since in general the states are used in some form of feedback and the number of measurable states is usually much less than the actual number of state variables, limited by the availability of cheap and rugged sensors and the ease of installation. Furthermore the measurements from the existing sensors may be corrupted by significant noise. In the present paper both the stochastic and the deterministic cases are investigated for application to optimal control of ram velocity in injection molding.