We introduce a continuous-discrete robust Kalman filter, obtained by transforming the discrete-time procedure presented in [1], and apply it to reduce the effect of network delay in the path between state estimator and controller. Our continuous-discrete robust Kalman filter relies on time-stamping to mitigate the destabilizing effects of network delay in the presence of significant model parameter uncertainty. We show that our method can achieve acceptable performance in the presence of both communication delay and model parameter uncertainty. We also demonstrate the reduced sensitivity of our robust estimator to such uncertainty as compared with a standard Kalman filter.