Texas Air Quality Study field campaigns took place in eastern Texas in August–October of 2000 and 2006. Several flights of NOAA and NCAR research aircraft were dedicated to characterizing anthropogenic emissions over Houston. We present results from an inverse modeling technique that uses three atmospheric transport models and these aircraft observations to assess and improve existing emission inventories. We used inverse modeling techniques to improve the spatial and temporal emissions' distribution of CO, NOy, and SO2 predicted by the 4 km resolution U.S. Environmental Protection Agency (EPA) National Emission Inventory (NEI) for 2005. Differences between the prior and posterior inventories are discussed in detail. In September 2006, we found that the prior daytime CO emissions in the Houston urban area have to be reduced by 41% ± 8%. Over the Houston Ship Channel, where industrial emissions are predominant, the prior emissions have to be decreased by 43% ± 5% for CO and 51% ± 5% for NOy. Prior NOy emissions from other major ports around Houston also have to be reduced, probably owing to uncertain nearshore ship emissions in the EPA NEI inventory. Using the measurements from the two field campaigns, we assessed the emissions' variability between August 2000 and September 2006. Daytime CO emissions from the Houston urban area have decreased by 8% ± 3%, while the NOy emissions have increased by 20% ± 6%. In the Houston Ship Channel, daytime NOy emissions have increased by 13% ± 7%. Our results show qualitative consistencies with known changes in Houston emissions' sources.