Particulate matter is one of major pollutants causing the ambient air quality standard to be exceeded in southern Taiwan, especially during November to February in the following year. The results from field measurements showed that the abundant species in PM2.5 were sulfate, nitrate, ammonia, organic carbon and elemental carbon. In this study, Models-3/CMAQ was used to simulate the particulate matter in southern Taiwan from the 1st to 5th, November 2005 and the simulated results were then compared with field measured data. The input meteorological data for Models-3/CMAQ simulation were provided by MM5 with FDDA and the emission data were updated from Taiwan Emission Data System. Four layers of nested grids with the finest grid of 3 km by 3 km, 15 vertical sigma layers, RADM2 and CMAQv4.4 were used in the simulation. The simulated and measured daily mean concentrations for PM were in good agreement with the mean paired normalized error about 8%. Similar good agreements were also obtained for ozone, NOx, and SO2. Significant differences were found for the simulated and measured hourly concentrations for various particular compositions, although those for ozone, HNO3, and H2O2 were in good agreement. For secondary nitrate and sulfate, simulated nitrate concentrations were only overestimated at night by factors of 2 to 4 while simulated sulfate concentrations were underestimated by factor 2 to 10. The simulated OC concentrations were generally underestimated and those for EC concentrations were in fair agreement. The formation mechanisms for nitrate and sulfate need to be modified for simulation in southern Taiwan, especially for sulfate.