Conventional analyses of exploration geochemical data assume that the bank-ground is a constant or slowly changing value, equivalent to a plane or smoothlycurved surface. Nevertheless, It is better to regard the geochemical backgroundas a rugged surface, varying with changes in geology and environment. Thisregged surface can be estimated from observed geological, geochemical, andenvironmental properties by using robust multivariate statistics.Ametbod of background adjustment was developed and applied to ground-water and stream sediment reconnaissance data collected from the Black Hillsarea, South Dakota, U. S. A. The Iithology of source rocks seems to be a dominantfactor determining chemical composition of groundwater or stream sadiments. Themost efficacious adjustment procedure is to regress uranium comcentration on sele-cted geochemical and environmental variables for each lithologic unit, and thento delineate anomalies of the combined residuals. This is called the "methed ofunit-wise adjustment of geochemical background". Robust versions of regressionand RQ-mode principal components analyds rather than ordinary techniques wereused to guard against distortion caused by outliers. Anomalies delineated by thisbackground adjustment procedure correspond with uranium prospects much betterthan anomalies delineated by conventional procedures do. The procedure should beapplicable to geochemical exploration for other metals at different scales, andthe adoption of this procedure may result in a revolution in geochemical dataanalyses.