This paper qualifies the effect of different convection schemes and center-domains on the simulation of winter precipitation over Iran using RegCM3 numerical climate model. It is compressible, finite difference with hydrostatic equilibrium in sigma coordinate with a semi-implicit algorithm for reduction of horizontal diffusion. Iran experienced two dry and wet periods during winters of 1997 and 2000. In this paper sensitivity of the RegCM3 regional climate model has been analyzed using different convection schemes, including Grell-AS, Grell- FC, Emanuel and Kuo and different center domains including Himalaya, Mediterranean sea, Iran and Indian ocean (near Pakistan southern boundaries). NNRP1 data with 5 degree resolution, GLCC and GTOPO are used as initial and boundary conditions, land use and topography data, respectively. We found that there is an intelligence between model skill and center domain, for example Himalaya center domain gives good results when a Siberian high pressure develops over Iran. The minimum bias and MAE of -0.7mm and 0.6mm are captured by Kuo convection scheme in the modeling of dry year 2000, but Grell convection scheme have minimum error among two wet and dry winter simulation. Precipitation amounts are underestimated for both convection and center domain experiments when compared to observations during winter 2000. The largest difference occurred in Himalaya experiment. We found that CRU reanalysis data can not be considered as a reference data for calibration of RegCM3 over Iran. It is also found that there is an eastward shift in Caspian Sea precipitation pattern in all experiments.