Abstract: Using classical steady-state Kalman filtering theory, a new approach of designing Wiener state estimators is presented, whose principle is that based on steady-state Kalman filter and predictor given in the Wiener filter form, and using the autoregressive moving average(ARMA) innovation model, the recursive version of non-recursive steady-state optimal state estimators yields the Wiener state estimators. The proposed Wiener state estimators can handle the state filtering, smoothing and prediction problems in a unified framework. They have the ARMA recursive form, and have asymptotic stability and optimality. A simulation example shows their effectiveness.