Soil dust aerosol is the largest contributor to aerosol mass concentrations in the troposphere and has considerable effects on air quality and climate. Arid and semi-arid areas of East Asia are one of the important dust source regions thus it is crucial to understand dust mobilization and accurately estimate dust emissions in East Asia. However, present dust models still contain large uncertainties with dust emissions that remain a significant contributor to the overall uncertainties in the model. In this study, we attempt to reduce these uncertainties by using an inverse modeling technique and obtain optimized dust emissions. We use Moderate Resolution Imaging Spectrometer (MODIS) aerosol optical depths (AODs) and groundbased mass concentrations of particles less than 10 μm in aerodynamic diameter (PM10) observations over East Asia in May 2007. The MODIS AODs are validated with AErosol RObotic NETwork (AERONET) AODs. The inversion uses the maximum a posteriori method and the GEOS-Chem chemical transport model (CTM) as a forward model. The model error is large over dust source regions including the Gobi Desert and Mongolia. We find that inverse modeling analyses from the MODIS and PM10 observations consistently result in decrease of dust emissions over Mongolia and the Gobi Desert. Whereas over the Taklamakan Desert and Manchuria, the inverse modeling analyses from both observations yield contrast results such as increase of dust sources using MODIS AODs, while decrease of those using PM10 observations. We discuss some limitations of both observations to obtain the optimized dust emissions and suggest several strategies for the improvement of dust emission estimates in the model.