We collected mercury observations as part of the Nitrogen, Oxidants, Mercury, and Aerosol Distributions, Sources, and Sinks (NOMADSS) aircraft campaign over the southeastern US between 1 June and 15 July 2013. We use the GEOS-Chem chemical transport model to interpret these observations and place new constraints on bromine radical initiated mercury oxidation chemistry in the free troposphere. We find that the model reproduces the observed mean concentration of total atmospheric mercury (THg) (observations: 1.49 ± 0.16 ng m⁻³, model: 1.51 ± 0.08 ng m⁻³), as well as the vertical profile of THg. The majority (65 %) of observations of oxidized mercury (Hg(II)) were below the instrument's detection limit (detection limit per flight: 58–228 pg m⁻³), consistent with model-calculated Hg(II) concentrations of 0–196 pg m⁻³. However, for observations above the detection limit we find that modeled Hg(II) concentrations are a factor of 3 too low (observations: 212 ± 112 pg m⁻³, model: 67 ± 44 pg m⁻³). The highest Hg(II) concentrations, 300–680 pg m⁻³, were observed in dry (RH  <  35 %) and clean air masses during two flights over Texas at 5–7 km altitude and off the North Carolina coast at 1–3 km. The GEOS-Chem model, back trajectories and observed chemical tracers for these air masses indicate subsidence and transport from the upper and middle troposphere of the subtropical anticyclones, where fast oxidation of elemental mercury (Hg(0)) to Hg(II) and lack of Hg(II) removal lead to efficient accumulation of Hg(II). We hypothesize that the most likely explanation for the model bias is a systematic underestimate of the Hg(0) + Br reaction rate. We find that sensitivity simulations with tripled bromine radical concentrations or a faster oxidation rate constant for Hg(0) + Br, result in 1.5–2 times higher modeled Hg(II) concentrations and improved agreement with the observations. The modeled tropospheric lifetime of Hg(0) against oxidation to Hg(II) decreases from 5 months in the base simulation to 2.8–1.2 months in our sensitivity simulations. In order to maintain the modeled global burden of THg, we need to increase the in-cloud reduction of Hg(II), thus leading to faster chemical cycling between Hg(0) and Hg(II). Observations and model results for the NOMADSS campaign suggest that the subtropical anticyclones are significant global sources of Hg(II).