Numerous sites in the United States and around the world are contaminated with depleted uranium (DU) in various forms. A prevalent form is fragmented DU originating from various scientific tests involving high explosives and DU during weapon-development programs, at firing practice ranges, or in war theaters where DU was used in armor-piercing projectiles. The contamination at these sites is typically very heterogeneous, with discrete, visually identifiable DU fragments mixed with native soil. The bulk-averaged DU activity is quite low, whereas DU fragments, which are distinct from the soil matrix, have much higher specific activity. DU is best known as a dark metal that is nearly twice as dense as lead, but DU in the environment readily weathers (oxidizes) to a distinctive bright yellow color that is quite visible. While the specific activity (amount of radioactivity per mass of soil) of DU is relatively low and presents only a minor radiological hazard, the fact that DU is radioactive and visually identifiable makes it desirable to remove the DU ''contamination'' from the environment. The typical approach to conducting this DU remediation is to use radiation-detection instruments to identify the contaminant and then to separate it from the adjacent soil, packaging it for disposal as radioactive waste. This process can be performed manually or by specialized, automated equipment. Alternatively, a more cost-effective approach might be simple mechanical or gravimetric separation of the DU fragments from the host soil matrix. At SNL/NM, both the automated and simple mechanical approaches have recently been employed. This paper discusses the pros/cons of the two approaches.