This paper is concerned with the optimal measurement feedback control and estimation problem for network control systems (NCSs) with state-package dropouts. It is noted that the considered problem in this paper is essentially the measurement feedback control for a multiplicative-noise stochastic system. The key contribution of this paper is to derive the optimal estimator (conditional expectation) for NCSs with state-package dropouts. Different from the intermittent Kalman filter and the linear minimum mean square error estimator, the presented estimator is proved to be optimal without conditioning on the arrival process {γ_κ}. With the use of maximum principle, the optimal measurement feedback control and the optimal estimator can be solved separately, and the optimal performance index is achieved under the optimal control strategy.