The Japan Meteorological Agency (JMA) conducted the second Japanese global atmospheric reanalysis, called JRA-55, covering the period from 1958. JRA-55 is the first comprehensive reanalysis to cover the last half-century since the European Centre for Medium-Range Weather Forecasts 45-year Reanalysis (ERA-40) and the first to apply four-dimensional variational analysis (4D-Var) to this period. The main objectives of JRA-55 were to address issues found in previous reanalyses and to produce a comprehensive atmospheric dataset suitable for studying multidecadal variability and climate change. JRA-55 was produced using the TL319 version of JMA's operational data assimilation system, which was extensively improved since the Japanese 25-year Reanalysis (JRA-25). It also uses several newly available and improved past observations. The resulting reanalysis products are considerably better than the JRA-25 product. Two major problems of JRA-25 were a cold bias in the lower stratosphere, which has been diminished, and a dry bias in the Amazon basin, which has been mitigated. The temporal consistency of temperature analysis has also been considerably improved compared to previous reanalysis products. Initial quality evaluation revealed problems such as a warm bias in the upper troposphere, large upward imbalance in the global mean net energy fluxes at the top of the atmosphere and at the surface, excessive precipitation over the tropics, and unrealistic trends in analyzed tropical cyclone strength. The paper also assesses the impacts of model biases and changes in the observing system, and mentions efforts to further investigate the representation of low-frequency variability and trends in JRA-55. JRA-55 uses a wide range of observational data, including conventional data, satellite radiances, microwave sounders, microwave imagers, and GNSS-RO refractivities. The data assimilation system used in JRA-55 is based on the low-resolution (TL319) version of JMA's operational data assimilation system as of December 2009. The forecast model used for JRA-55 is based on the TL319 spectral resolution version of the JMA global spectral model (GSM) as of December 2009, which has been extensively improved since JRA-25. The dynamics of the forecast model include a reduced Gaussian grid to mitigate the overconcentration of grid points at high latitudes and lower the computational cost. The radiation calculations in the forecast model have been improved, with the longwave radiation fluxes computed by solving the equation of radiative transfer for a non-scattering atmosphere in each predefined spectral band. The shortwave radiation fluxes are computed by a two-stream method with the δ-Eddington approximation. The cloud radiation calculations in the forecast model treat clouds as blackbodies and assume maximumThe Japan Meteorological Agency (JMA) conducted the second Japanese global atmospheric reanalysis, called JRA-55, covering the period from 1958. JRA-55 is the first comprehensive reanalysis to cover the last half-century since the European Centre for Medium-Range Weather Forecasts 45-year Reanalysis (ERA-40) and the first to apply four-dimensional variational analysis (4D-Var) to this period. The main objectives of JRA-55 were to address issues found in previous reanalyses and to produce a comprehensive atmospheric dataset suitable for studying multidecadal variability and climate change. JRA-55 was produced using the TL319 version of JMA's operational data assimilation system, which was extensively improved since the Japanese 25-year Reanalysis (JRA-25). It also uses several newly available and improved past observations. The resulting reanalysis products are considerably better than the JRA-25 product. Two major problems of JRA-25 were a cold bias in the lower stratosphere, which has been diminished, and a dry bias in the Amazon basin, which has been mitigated. The temporal consistency of temperature analysis has also been considerably improved compared to previous reanalysis products. Initial quality evaluation revealed problems such as a warm bias in the upper troposphere, large upward imbalance in the global mean net energy fluxes at the top of the atmosphere and at the surface, excessive precipitation over the tropics, and unrealistic trends in analyzed tropical cyclone strength. The paper also assesses the impacts of model biases and changes in the observing system, and mentions efforts to further investigate the representation of low-frequency variability and trends in JRA-55. JRA-55 uses a wide range of observational data, including conventional data, satellite radiances, microwave sounders, microwave imagers, and GNSS-RO refractivities. The data assimilation system used in JRA-55 is based on the low-resolution (TL319) version of JMA's operational data assimilation system as of December 2009. The forecast model used for JRA-55 is based on the TL319 spectral resolution version of the JMA global spectral model (GSM) as of December 2009, which has been extensively improved since JRA-25. The dynamics of the forecast model include a reduced Gaussian grid to mitigate the overconcentration of grid points at high latitudes and lower the computational cost. The radiation calculations in the forecast model have been improved, with the longwave radiation fluxes computed by solving the equation of radiative transfer for a non-scattering atmosphere in each predefined spectral band. The shortwave radiation fluxes are computed by a two-stream method with the δ-Eddington approximation. The cloud radiation calculations in the forecast model treat clouds as blackbodies and assume maximum