A 3D-VAR analysis was developed with the intention of operational application in the Rapid Update Cycle (RUC) at NCEP. In the 3D-VAR method we combine background fields with observations from different sources in an optimal way. The main sources of observations are RAOBs, METARs, buoys, aircrafts, wind profilers, boundary layer profilers, radar (VAD) winds, satellites, and GPS precipitable retrievals. New observation error terms are applied and a new, better balanced background term was introduced. The background error correlations are applied through a linear combination of digital pseudo-Gaussian filters, which approximates SOAR-type correlations. The observation error variances guarantee good fit to the observations. This ensures an accurate reproduction of the observed atmospheric structures, which is essential for nowcasting type applications of the analysis. Analysis corrections are applied in a variation of quasi-isentropic coordinate used in the RUC forecast model. This structure is, of course, adaptive in space and time. The 3D-VAR scheme is tested using idealized observations and also using actual observations in a real time cycled environment at the NOAA Forecast Systems Laboratory. The description of the analysis method and its testing results will be presented. The implementation of 3dvar into the RUC has important consequences for RUC forecast users regarding fit to observations and 3d-structures. It also sets the stage for future assimilation of radar reflectivity and radial winds, and satellite radiances.