With the increase of computer power, the skill of medium range numerical weather prediction has been improved. The improvement of forecast skill in the forecast model system can be attributed to the improved initial condition, model dynamics and physics (Kalnay 2003). Meanwhile, there have been relatively energetic research activities on the development of physics algorithms in research community, which has mainly evaluated for short-range forecasts and climate simulations.These sophisticated physics schemes have been evaluated on the short range forecast for severe weather in regional model or climate simulations, but not on the medium-range forecast. It is thus important to investigate the impacts of physics processes on the medium-range forecast. This study aims to investigate the role of a revised physics package against the old package of a global model in medium range forecasts. A version of NCEP GSM is used. The sensitivity of the package to the vertical resolution is also investigated. Verifications against initial data and observations are also performed for 10 day forecasts in August 2010 and January 2011. An additional set of experiments for a heavy rainfall case over Korea is carried out in order to examine the individual role of each physics scheme in improving the forecast skill.