This study was carried out to find the optimal methods in odor dispersion modeling. The CALMET and CALPUFF recommended by US EPA were used in the study. The accuracy of 3 dimensional meteorological field was one of the important parameters in the modeling, To understand the sensitivity of CALMET according to meteorological input data, four Cases were tested and compared with the measurements. The four Cases with various input data were followed : Case 1 was surface and upper measured data, Case 2 was MM5 data only, Case 3 was surface weather station and MM5 data, Case 4 was surface weather station, upper-air weather station and MM5 data. The comparisons of wind speeds, wind direction and temperatures by CALMET model with observations under various input data showed that Case 4 was more accurate than the other Cases. The results of CALPUFF dispersion modeling were compared with odor complains data and they showed that there were similar patterns even if some discrepancies existed. CALPUFF dispersion modeling was also compared with odor monitoring station data, and there was a good agreement in the threshold odor concentrations. In order to get more accurate result for odor modeling, we need following additional measures: First, fine tuning of odor emission data with the consideration of variations of emission fate according to operation hour of the odor emitting facilities. Second, construction of continuous monitoring system(on-line GC, odor a sensor etc.) was necessary to monitor the odor compound concentrations.