Ozone pollution in the Southeast US involves complex chemistry driven by emissions of anthropogenic nitrogen oxide radicals (NOx ≡ NO + NO2) and biogenic isoprene. Model estimates of surface ozone concentrations tend to be biased high in the region and this is of concern for designing effective emission control strategies to meet air quality standards. We use detailed chemical observations from the SEAC4RS aircraft campaign in August and September 2013, interpreted with the GEOS-Chem chemical transport model (CTM) at 0.25°×0.3125° horizontal resolution, to better understand the factors controlling surface ozone in the Southeast US. We find that the National Emission Inventory (NEI) for NOx from the US Environmental Protection Agency (EPA) is too high in the Southeast and nationally by 50 %. This is demonstrated by SEAC4RS observations of NOx and its oxidation products, by surface network observations of nitrate wet deposition fluxes, and by OMI satellite observations of tropospheric NO2 columns. Upper tropospheric NO2 from lightning makes a large contribution to the satellite observations that must be accounted for when using these data to estimate surface NOx emissions. Aircraft observations of upper tropospheric NO2 are higher than simulated by GEOS-Chem or expected from NO-NO2-O3 photochemical stationary state. NOx levels in the Southeast US are sufficiently low that only half of isoprene oxidation proceeds by the high-NOx pathway to produce ozone; this fraction is only moderately sensitive to changes in NOx emissions because isoprene and NOx emissions are spatially segregated. GEOS-Chem with reduced NOx emissions provides an unbiased simulation of ozone observations from the aircraft and from ozonesondes, and reproduces the observed ozone production efficiency in the boundary layer as derived from a regression of ozone and NOx oxidation products. However, the model is still biased high by 8 ± 13 ppb relative to observed surface ozone in the Southeast US. Ozonesondes launched during midday hours show a 7 ppb ozone decrease from 1.5 km to 0.2 km altitude, whereas GEOS-Chem has no such gradient because of efficient boundary layer mixing. We conclude that model biases in simulating surface ozone over the Southeast US may be due to a combination of excessive NOx emissions and excessive boundary layer vertical mixing.