Resuspended dust from paved roads in Seoul and Incheon metropolitan areas is regarded as one of the major $PM_{10}$ sources in these areas, according to the recent emission estimates using the emission factors compiled in AP-42. It is well known that the AP-42 model for estimating $PM_{10}$ emissions from paved roads requires information on silt loadings of particular paved roads. The conventional AP-42 method (vacuum swept method) for road silt sampling, however, is expensive, time consuming, and dangerous. These drawbacks led us to develop a Mobile Dust Monitoring System (MDMS) capable of doing real time measurements of silt loading of paved roads, thereby we could get higher resolution silt loading data both in terms of time and space without too much human efforts and danger. In this study, for the real-time measurement of silt loading of paved roads, the principle used in the TRAKER method of U.S. Desert Research Institute was employed and the entire sampling systems including data acquisition system were designed for theses purpose and mounted on a SUV. The correlation between the silt loading measured by vacuum swept method and the speed corrected ${Delta}Dust$ was derived for the vehicle-based silt loading measurements, and then the variations of silt loading on paved roads were surveyed using the MDMS in test routes of Seoul and Incheon. From the results of real-time measurements, temporal and spatial variations of silt loading data together with the existence of hot spots were observed for paved roads in Seoul and Incheon. The result of this study will be employed to estimate fugitive dust emissions from paved roads.