Atmospheric deposition represents a major input of mercury to surface environments. The phase of mercury (gas or particle) has important implications for its removal from the atmosphere. We use long-term observations of reactive gaseous mercury (RGM), particle-bound mercury (PBM), fine particulate matter (PM2.5), and temperature at five sites in North America to derive an empirical gas-particle partitioning relationship log10(K-1) = (10 ± 1) − (2500 ± 300)/T where K = (PBM/PM2.5)/RGM with PBM and RGM in common mixing ratio units, PM2.5 in μg m−3, and T in Kelvin. This relationship is in the range of previous work but is based on far more extensive data from multiple sites. We implement this empirical relationship in the GEOS-Chem global 3-D Hg model to partition divalent mercury (Hg(II)). The resulting gas-phase fraction of Hg(II) ranges from over 90% in warm air with little aerosol to less than 10% in cold air with high aerosol. Hg deposition to high latitudes increases because of more efficient scavenging of particulate Hg(II) by snow. Model comparison to Hg observations at surface sites suggests that subsidence from the free troposphere (warm air, low aerosol) is a major factor driving the seasonality of RGM, while elevated PBM is mostly associated with high aerosol loads. This and other model updates, including the correction of an outstanding algorithm error, to wet deposition improve the simulation of Hg wet deposition fluxes in the US relative to the previous version of the model. The observed wintertime minimum in wet deposition fluxes is attributed to inefficient snow scavenging of gas-phase Hg(II).