The operational DA technology in use at the Russian HydroMetCenter is outlined. A new 3D-Var scheme under development is presented. The scheme is intended to be applicable for data assimilation on all spatial scales (global, regional, and meso-scale). The core of the scheme can be used for ocean data assimilation. The analysis scheme is designed to be flexible enough to efficiently utilize spatially variable flow-dependent background-error covariances expected from a future Ensemble Kalman Filter. The new 3D-Var scheme relies on 3-D physical-space spatial filters. Capability of the covariance model to represent spatially variable structures is demonstrated. Ongoing research in satellite data assimilation is sketched. Current status and prospects of the operational deterministic medium-range weather forecasting at HMC is overviewed. Ensemble prediction issues at HMC are presented. A recent study of the spatio-temporal structure of model errors is briefly presented. Issues of possible NOAA-RosHydromet collaboration are discussed.