The need by land managers and planners for more quantitative measures of mineral values has prompted scientists at the U.S. Geological Survey to test a probabilistic method of mineral resource assessment on a portion of the wilderness lands that have been studied by the Survey during the past 20 years. A quantitative estimate of undiscovered mineral resources is made by linking the techniques of subjective estimation, geologic mineral deposit models, and Monte Carlo simulation. The study, which uses grade-tonnage and occurrence models for 21 geologic deposit types, considers 91 U.S. Forest Service wilderness tracts in California, Nevada, Oregon, and Washington. Estimates of the amounts of the 11 metals contained in undiscovered mineral deposits of the types studied range from negligible to several years of U.S. consumption. Although these estimates are limited to metals contained in undiscovered deposits of a small number of metallic mineral deposit types, the assessment procedure can be expanded by the use of additional deposit models and by using information about identified mineral resources. This will allow models of economic processes such as exploration, development, and production to be applied.