Due to climate changes, precipitation forecast average in time scale is one of the most important challenges for specialists in the recent years. The purpose of this research is to investigate the capabilities of the dynamic model RegCM4 in precipitation forecast in cold period of Fars province. In this study, September to February or 6 month is considered as the cold period. Several variable statistical periods 1990-2010 are selected. In this study, two data sets are used by post-processing methods using statistical regression techniques. 1 - Data needed for implementation of Dynamic Model RegCM4 was taken from the Centre ICTP with a NetCDF format including weather data on a daily scale (6 hours) with a horizontal grid 5/2 × 5/2 degrees, sea level data, with 1 ° grid surface data. 2 - Monthly precipitation data (mm) watch of seven synoptic stations which have been received from the Meteorological province. In order to implement dynamic model, the test of scheme determination is performed and investigation showed that Darrell’ scheme in comparison with the two other schemes, Koo and Emmanuel, has fewer errors in the modeling the rainfall during the cold season in Fars stations. After running the model, outputs were processed using multivariate regression methods. In order to enhance the efficiency of RegCM4 model, 20 km horizontal resolution model output (dynamic) using multiple regression the statistical post-processing (dynamic-data) groups. Double precipitation data with a resolution of 20 km and precipitation observations of monthly precipitation data were compared with the post-processed to determine the performance of the statistical processing on the output RegCM4 model.The results with comparing the data showed that in 43% of the stations in autumn the use of raw output of climate model precipitation RegCM4 and dynamic-statistic method have had the same efficiencies and in about 14% of cases neither of the two options have preference to the other. In winter many more stations represent the efficiency of using the raw climate model RegCM4 as 4 out of 7 stations have confirmed the superiority of the model in this season (57 percent) while the number of successful stations using the dynamic-Statistics is 2 (29 percent). Also 1 station (14%) in applying the above two cases do not have a specific preference. In the cold period of Fars Province, the number of the stations adapted with raw output of climate model (rainfall) RegCM4, and the output of dynamic-statistic method is respectively 5 cases (71%) and 2 cases (29%). in the study of rainfall in cold period no cases has been found that none of the options is not superior to the other in it. In general we can say that in 1.57% of cases the output of RegCM4 model and in 3.33% of cases the output of dynamic-statistic method has better ability to predict rainfall of Fars in cold period.Therefore, we can conclude that the small-scale dynamic view of 20 × 20 km horizontal resolution needed to apply statistical post-processing or dynamic-data to enhance the accuracy of the data is not mentioned.