Precipitation runoff experiments were conducted on copper panels at sites polluted with sulfur dioxide (Washington, DC), an unpolluted site (Albany, OR), and an unpolluted marine site (Newport, OR). Copper patinas had complex surface chemistry representing a sequence of transitional mineral phases terminating in brochantite at the polluted sites, atacamite at the unpolluted marine site, and copper oxides at the unpolluted site. Precipitation runoff rates on large panels were 3.3 (0.37), 1.7 (0.19), and 1.7 (0.19) g Cu/m2y (μm Cu/y) for the Washington, DC, Albany, OR, and Newport, OR, sites, respectively. Higher rates were observed for small panels and appeared due to the reduced impact of residence time effects. Zinc corrosion products on the panel at Newport depressed the copper runoff until the zinc corrosion products were dissipated. Copper runoff at the Washington, DC, site was partitioned into contributions from dry deposition of acidic gases, and wet deposition of strong acids and weak acid (carbonic acid) using geochemical modeling software. The software also predicted precipitation of brochantite during winter when sulfur dioxide levels were highest. An empirical atmospheric corrosion model with early parabolic corrosion kinetics for copper and long-term linear kinetics, and including synergy between properties of the copper patina and the environment, was described. Applied to the patina on a 100-year old roof, the model suggests that only 15 % of the total mass loss is retained in the corrosion film after 100 years of exposure.