Improvements to the NCAR Climate System Model, CSM-1, primarily for transient climate forcing simulations, are discussed. The impact of the individual changes is assessed through atmosphere–land or ocean–ice experiments, and through short coupled simulations. A 270-yr control simulation has been performed using the model with all of the changes, defined as CSM-1.3. Trace gas concentrations appropriate for 1870 were used and the model produced a stable surface climate with less deep ocean drift than CSM-1. Changing the aerodynamic roughness length of sea ice to a value appropriate for first year ice reduced the deep ocean salinity trend by a factor of 100 and error in the transport by the Antarctic circumpolar current by 50% (60 × 106 m3 s−1). Three new features were added to the Community Climate Model, version 3 (CCM3), which is the atmospheric component of CSM-1. These additions were N2O, CH4, CFC11, and CFC12 as prognostic tracers and the oxidation of CH4 to form water vapor; a prognostic clou... Abstract Improvements to the NCAR Climate System Model, CSM-1, primarily for transient climate forcing simulations, are discussed. The impact of the individual changes is assessed through atmosphere–land or ocean–ice experiments, and through short coupled simulations. A 270-yr control simulation has been performed using the model with all of the changes, defined as CSM-1.3. Trace gas concentrations appropriate for 1870 were used and the model produced a stable surface climate with less deep ocean drift than CSM-1. Changing the aerodynamic roughness length of sea ice to a value appropriate for first year ice reduced the deep ocean salinity trend by a factor of 100 and error in the transport by the Antarctic circumpolar current by 50% (60 × 106 m3 s−1). Three new features were added to the Community Climate Model, version 3 (CCM3), which is the atmospheric component of CSM-1. These additions were N2O, CH4, CFC11, and CFC12 as prognostic tracers and the oxidation of CH4 to form water vapor; a prognostic clou...