Particulate matter (PM) pollution has been of concern owing to its adverse effects on human health as well as its role in visibility degradation by atmospheric haze. Such quality of life concerns have resulted in an increased interest in the analysis and prediction of urban PM. Windblown dust is one of the important PM source categories in arid/semiarid regions and is often poorly characterized in the existing air quality models. Our previous study, modeling episodic high PM events along the U.S./Mexico border using the sophisticated CMAQ/MM5/SMOKE air quality modeling system, identified that some of the observed PM10 NAAQS exceedances over the semi-arid area were closely related to windblown dust and the modeling system was incapable of accounting for the time-dependent episodic dust emission flux. In this study, efforts were made to implement a time-dependent windblown dust flux into the CMAQ/MM5/SMOKE modeling system with the hope of improving the predictability of the PM concentration in semiarid/arid regions. An approach for parameterizing hourly windblown dust flux for each grid cell over the study domain is presented in detail along with comparisons of model predictions with the observations for the case of high PM events in a pair of U.S./Mexico border towns. The results showed an improved predictability of PM10 concentrations when the windblown dust emission flux, which is a function of the soil moisture content, soil type, atmospheric stability, and wind speed, is accounted for in model formulation.