This paper presents an optimization-based algorithm for state estimation of multirate digital control systems. The algorithm features a predictor estimator structure and it is derived by employing a time invariant modelling approach to multirate systems. Special care is taken to ensure the causality of the derived estimation algorithm and to reduce the dimensionality inherent in multirate system modelling while maintaining a reasonably simple implementation. The proposed algorithm is developed to ensure convergence in state estimation between base sampling instants as well as between model rate sampling instants.