Abstract Both the number of trees sampled per species and the number of sampling points under each tree are important throughfall sampling considerations. Chemical loadings obtained from an urban throughfall study were used to evaluate the relative importance of both of these sampling factors in tests for determining species' differences. Power curves for detecting differences among species derived from the noncentrality parameter developed herein indicated that the number of trees sampled per species affects power more than the number of points sampled under each tree. For. Sci. 35(1):173-182.