Multivariate regression analysis was used to determine the major emission sources contributing to suspended particulates in air collected over a small mixed land use watershed. Aerosol samples were collected at 5 different land use types in 1976 and 1977 and analyzed for 10 elements. Aerosol composition was then compared to emission source composition to obtain the relative contribution by each emission source. Two examples of source reconciliation by multivariate regression analysis for typical samples are given. An industrial land use type clearly shows increased contributions from anthropogenic activity when compared to a rural site only 15 km distant in a crosswind direction. Differences in calculated and measured ambient aerosol concentrations can lead to specific interpretations about emission sources. The implications of errors in concentration data are discussed. A weighting technique can aid in correcting problems caused by incomplete or inaccurate data for 1 or 2 elements.