The global and regional data assimilation systems operational at the National Centers for Environmental Prediction (NCEP) are now using the Gridpoint Statistical Interpolation (GSI) algorithm, which was implemented in June 2006 and May 2007 in the regional and global systems respectively. Although the GSI algorithm currently utilizes a three-dimensional variational (3-D Var) framework, the GSI code has many new capabilities embedded and under development. These capabilities will enable better use of high resolution observations in time and space, such as from the NPOESS advanced sounder suite, while continuing to remain affordable in NCEP's operational environment. In the one year time frame, testing will include the First-Order Time-extrapolation to Observations (FOTO) algorithm, which is a simplified extension of 3-D Var into the time dimension. In addition, the GSI has been extended to allow scaling of the background error variance by the time change of the background field over the analysis time window. In a longer time frame, algorithms to generate situation-dependent background error covariances are being added. Collaborations with the NASA Goddard Space Flight Center Global Modeling and Analysis Office (NASA/GMAO) are continuing to focus on a 4D-Var application for the GSI. Last, the NCEP Environmental Modeling Center is working with additional collaborators in the research community on Ensemble Kalman Filtering (EnKF) approaches using NCEP data assimilation and model applications. A broad-based strategy to investigate these diverse approaches will be summarized along with the latest testing results.