ABSTRACT The Colorado River and its tributaries drain a complex provenance that includes multiple tectonic elements of the Proterozoic to Phanerozoic Cordilleran margin. In order to provide (1) a more reliable estimate than that reported in the literature for the composition of Holocene Colorado River sand, and (2) a standard for comparison with ancient sandstones that may contain detritus derived from a similar complex provenance, 25 unconsolidated samples were collected along the lower reaches of the Colorado River for detailed petrological examination. Each specimen was artificially consolidated with epoxy, thin sectioned, and stained for feldspar. Each author, utilizing the Gazzi-Dickinson point-count method, counted 400 points per sample; the results from the two point-count operators wer averaged. The average Colorado River sand in our collection is medium-grained, moderately to well sorted, and composed of 66.2%Q, 18.1%F, 14.7%L, and 1.3% miscellaneous grains. Q, on average, is composed of 89% monocrystalline and 11% polycrystalline quartz grains. The average plagioclase-to-total-feldspar ratio is 0.66, and the average unstable-aphanitic-rock-fragment population (L) consists of 6% metamorphic (Lm), 43% volcanic (Lv), 51% sedimentary (Ls), and 1% alterite fragments. On the QFL provenance-discrimination diagram developed by W. R. Dickinson and associates, our point-count data plot entirely within the recycled-orogen field. W. R. Dickinson and colleagues ascribed such provenances to subduction complexes, collision orogens, and foreland uplifts ass ciated with fold-and-thrust belts. Rocks in the Colorado River drainage basin, however, include (1) remnants of Precambrian crystalline rocks which form a part of cratonal North America; (2) miogeoclinal and platformal sequences that developed along the North American continental margin during Paleozoic time; (3) deformed Paleozoic and Mesozoic sedimentary successions that make up the Sevier-Laramide fold-and-thrust belt; (4) Mesozoic and Cenozoic sedimentary fill of the Sevier foreland basin; (5) Cenozoic basement-cored uplifts that formed during the Laramide orogeny; (6) plutonic and volcanic rocks of the Mesozoic and Cenozoic Cordilleran continental-margin magmatic arc; and (7) Cenozoic sedimentary and volcanic rocks associated with extension and dextral shear along the Cordilleran co tinental margin. Thus, the QFL compositions of Holocene sand derived from the Colorado River drainage basin do not fit neatly into the recycled-orogen model proposed by W. R. Dickinson and colleagues. In contrast, on the QmFLt provenance-discrimination diagram also engendered by W. R. Dickinson and associates, 16 samples plot in the mixed-provenance field, and nine specimens plot in the quartzose-recycled or transitional-continental fields. On the LmLvLs diagram formulated by R. V. Ingersoll and C. A. Suczek, our samples plot close to the Lv-Ls join, approximately midway between the fields of magmatic arcs, rifted continental margins, and suture belts, probably as a result of their mixed heritage. Thus, the QmFL SUB>t and LmLvLs diagrams produce results that reflect more accurately, than do the results derived from the QFL provenance-discrimination diagram, the complex multiple tectonic characteristics of the source rocks in the Colorado River drainage basin. Our samples, therefore, represent excellent examples of mixed-provenance sands. Casual use of only the QFL diagram can lead to erroneous interpretations of provenances composed of more than one major tectonic element.