Weathering steel samples were exposed for periods of up to 30 months at nine air monitoring sites in the St. Louis, Missouri area. Climatic and air quality data were recorded during the exposure period and subjected to a rigorous evaluation to eliminate recording errors and to estimate missing values. Weight loss was used as the measure of steel corrosion. Corrosion rate was evaluated with respect to, (1) flux of pollutants to the steel during both wet and dry periods. (2) temperature, and (3) exposure history. Different definitions of when the steel was wet were evaluated to determine the most likely 'critical relative humidity.' Non-linear multiple regression techniques were used to determine the statistical significance of each factor and develop a theoretically consistent environmental damage function.