A study of methods of collecting and processing samples to determine whether or not gold is present in areas of moderate size was undertaken in the northwestern part of the San Juan Mountains, Colo. As part of this study, 57 samples of pan concentrates were taken from streams draining three types of areas: (1)'barren' areas, where gold mineralization might be geologically possible but no deposits are known, (2) slightly mineralized areas that contain only a few known veins and prospects and small mines, and (3) well-mineralized areas that contain numerous veins and some very productive mines. The concentrate samples were analyzed by the fire-assay-atomic-absorption method. Replicate analyses of large samples gave results consistent enough to permit placing considerable confidence in the results obtained for smaller samples on which only one analysis was made. For general field practice, it is necessary to pan only enough sand and gravel to yield about 15 grams of concentrate. The analytical r6sults are also quantitatively compatible with known geologic relations and indicate that a few samples from a stream are adequate to distinguish between 'barren' and mineralized areas and to determine the relative amount of gold in mineralized areas. In other areas of similar gold deposits, data of this type should help decide whether more intensive search for gold deposits is justified.