Analytical and statistical analyses of data obtained from a contracted exposure study in St. Louis, Missouri, reveal that many microclimate differences can account for observed differences in corrosion behavior within a geographic region. Data from Mansfeld's Atmospheric Corrosion Monitors (ACMs) were evaluated with data from the Regional Air Monitoring System (RAMS). Relative humidity, temperature, windspeed, and levels of total sulfur gases as SO2 and oxides of nitrogen as NO2 were found to be statistically significant variables. Problems of covariance were avoided by partitioning the large data set into subsets. Relative humidity was found to be the most important but least accurate variable. Because it is seldom measured at exposure sites, an equation was developed to relate site to site relative humidity differences to temperature differences. With average relative humidity and another empirical equation, time-of-wetness can be estimated. The results are in good agreement with time-of-wetness from the ACMs.