The paper describes a new operational global sea surface temperature (SST) analysis produced by the National Meteorological Center (NMC) of NOAA. The analysis uses 7-day in situ (ship and buoy) and satellite SST data, processed weekly and daily using optimum interpolation (OI) on a 1° grid. The OI technique requires the specification of data and analysis error statistics, which are derived and show that ship data errors are nearly twice as large as those from buoys or satellites. The average e-folding spatial error scales are 850 km in the zonal direction and 615 km in the meridional direction. The analysis includes a preliminary step to correct satellite biases relative to in situ data using Poisson's equation, demonstrating its importance using data following the 1991 Mt. Pinatubo eruptions. The OI analysis has been computed using in situ and bias-corrected satellite data from 1985 to the present. The paper discusses the SST data sources, including in situ observations from ships and buoys, and satellite observations from the Advanced Very High Resolution Radiometer (AVHRR) on NOAA polar-orbiting satellites. It also details the OI SST analysis process, including quality control procedures and the computation of weights for the analysis increments. The OI statistics, derived from a 1-year period (July 1990 to June 1991), show that the data errors are correlated and uncorrelated, with different characteristics for ship, buoy, and satellite data. The satellite data increments are partitioned into correlated and uncorrelated parts, and the guess error is defined as the local minimum of the satellite data increments. The paper compares the OI analysis with and without the new statistics, showing that the new statistics result in a smoother analysis with slightly larger root mean square (RMS) differences from the data. The OI analysis error is sensitive to the error statistics, with the new statistics leading to a slightly higher error. Finally, the paper discusses the OI bias correction, which uses Poisson's equation to correct for large-scale satellite biases, particularly those caused by stratospheric aerosols from volcanic eruptions. The correction is applied to both daytime and nighttime satellite retrievals, and its effectiveness is demonstrated using data from the 1991 Mt. Pinatubo eruption.The paper describes a new operational global sea surface temperature (SST) analysis produced by the National Meteorological Center (NMC) of NOAA. The analysis uses 7-day in situ (ship and buoy) and satellite SST data, processed weekly and daily using optimum interpolation (OI) on a 1° grid. The OI technique requires the specification of data and analysis error statistics, which are derived and show that ship data errors are nearly twice as large as those from buoys or satellites. The average e-folding spatial error scales are 850 km in the zonal direction and 615 km in the meridional direction. The analysis includes a preliminary step to correct satellite biases relative to in situ data using Poisson's equation, demonstrating its importance using data following the 1991 Mt. Pinatubo eruptions. The OI analysis has been computed using in situ and bias-corrected satellite data from 1985 to the present. The paper discusses the SST data sources, including in situ observations from ships and buoys, and satellite observations from the Advanced Very High Resolution Radiometer (AVHRR) on NOAA polar-orbiting satellites. It also details the OI SST analysis process, including quality control procedures and the computation of weights for the analysis increments. The OI statistics, derived from a 1-year period (July 1990 to June 1991), show that the data errors are correlated and uncorrelated, with different characteristics for ship, buoy, and satellite data. The satellite data increments are partitioned into correlated and uncorrelated parts, and the guess error is defined as the local minimum of the satellite data increments. The paper compares the OI analysis with and without the new statistics, showing that the new statistics result in a smoother analysis with slightly larger root mean square (RMS) differences from the data. The OI analysis error is sensitive to the error statistics, with the new statistics leading to a slightly higher error. Finally, the paper discusses the OI bias correction, which uses Poisson's equation to correct for large-scale satellite biases, particularly those caused by stratospheric aerosols from volcanic eruptions. The correction is applied to both daytime and nighttime satellite retrievals, and its effectiveness is demonstrated using data from the 1991 Mt. Pinatubo eruption.