Simulations of high southern latitudes with the high-resolution, limited-area Penn State/NCAR mesoscale model, version 4 (MM4), examine the impact of a moist physics parameterization and the success of a one-way nesting inside the global NCAR community climate model, version 2 (CCM2). Discretization is by 100 km resolution in the horizontal and 15 or 16 levels in the vertical. Initial and boundary conditions for the simulations are provided by analyses of the European Centre for Medium-Range Weather Forecasts or CCM2 simulations. Comparisons of dry and moist simulations of an austral winter month, June 1988, are used to examine the role of clouds in the regional meteorology. A cloud-free MM4 simulation with boundary conditions provided by CCM2 run 422 indicates that the one-way nesting of the mesoscale model can produce significant differences in the model output, including an improvement in the location of one longwave trough, reduced cold bias, and a more detailed surface wind field. The improved location of the longwave trough over the Atlantic Ocean is hypothesized to result from blocking induced by the reduced cold bias. The nested simulation is sensitive to the forcing at the horizontal boundaries. Consequently, proper location of troughs and ridges at the boundaries is required for the model to well represent all the major troughs and ridges inside the domain. The addition of moist physics to the mesoscale simulations tends to improve the quality of the simulated fields over the Southern Ocean. In particular, the intensity of the circumpolar trough is increased. Over Antarctica, serious deficiencies are found in the simulations with moist physics. Excessive moisture is apparently stored in the simulated clouds leading to excessive atmospheric back radiation and, consequently, excessive temperatures at the surface and higher up in the troposphere.¿ 1997 American Geophysical Union