Analysis of a common-garden study of broadleaf lupine (Lupinus latifolius Lindl. ex J.G. Agardh ssp. latifolius [Fabaceae]) indicates that use of watershed delineations is better than use of plant association series for determining seed zones on the Mt Hood National Forest. Risk analysis further confirmed that only 4 seed zones are required, providing a reasonable compromise between managing costs and maintaining local adaptation. Overall, moderate amounts of genetic variation were found in 84 seed sources. Two principal components (PCs) summarized 58% of the variation in 24 measured traits, and variation in PC scores was significantly correlated with topographic, geographic, and climatic variables. Regression analyses showed that these variables accounted for 47% of the variation in the first PC and 34% of the variation in the second PC.