Observations of tropospheric column densities of NO2 obtained from the Global Ozone Monitoring Experiment (GOME) for a 3-year period (1997, 1998, and 2000) are used to derive average seasonal variations in surface emissions of NOx from east China (100--1230E, 20--420N). The retrieval allows for zonal variations in the contribution of the stratosphere to the NO2 column and removes a bias of 110% on the seasonality of retrieved columns introduced by cloud screening. The top-down inventory is constructed using an inversion approach with a global 3-D chemical transport model (GEOS-Chem) and combined subsequently with the a priori inventory to develop an a posteriori inventory. The contribution of background NO2 arising from nonsurface sources (lightning) and long-range transport of emissions originating outside of east China is accounted for in the inversion. The a posteriori estimate of overall emissions for east China, 4.66 Tg N/yr (130% uncertainty), is 33% higher than the a priori value and is shown to improve agreement with surface measurements of nitrate wet deposition and concentrations of NOy observed in China. On the basis of multiple constraints on the spatial and seasonal variations of combustion and microbial processes, the a posteriori inventory is partitioned among emissions from biomass burning, fuel combustion, and microbial activity (or soil emissions). Emission of NOx from biomass burning in east China is estimated as 0.08 TgN/yr 1 50% in the a posteriori inventory, increased by about a factor of 2 from the a priori estimate. The resulting a posteriori inventory for fuel combustion (3.72 TgN/yr 1 32%) is about 15% higher than the a priori and exhibits a distinct maximum in winter, in contrast to the weak seasonality indicated in the a priori inventory. The a posteriori value for the microbial source of NOx (0.85 TgN/yr 1 40%) is about a factor of 3 higher than the a priori value, amounting to 23% of combustion sources for east China and significantly higher than a priori value of 7%. The microbial source is unimportant in winter. It peaks in summer, accounting for as much as 43% of the combustion source for that season, and is significant also in spring and fall. This seasonality is attributed to the timing of fertilizer application and to the influence of seasonally variable environmental factors including temperature and precipitation.