36Current advances in sequencing technology have greatly increased the availability of 37 sequence data from public genetic databases. With data from GenBank, we assemble and 38 phylogenetically investigate a 19,740-taxon, five-locus supermatrix (i.e., atpB, rbcL, matK, 39 matR, and ITS) for rosids, a large clade containing over 90,000 species, or approximately a 40 quarter of all angiosperms (assuming an estimate of 400,000 angiosperm species). The 41 topology and divergence times of the five-locus tree generally agree with previous estimates 42 of rosid phylogeny, and we recover greater resolution and support in several areas along the 43 rosid backbone, but with a few significant differences (e.g., the placement of the COM clade, 44 as well as Myrtales, Vitales, and Zygophyllales). Our five-locus phylogeny is the most 45 comprehensive DNA data set yet compiled for the rosid clade. Yet, even with 19,740 species, 46 current sampling represents only 16-22% of all rosids, and we also find evidence of strong 47 49 asterids, monocots) as well as other large, understudied branches of the Tree of Life, 50 highlighting the need for broader molecular sampling. Nevertheless, the phylogeny presented 51 here improves upon sampling by more than two-fold and will be an important resource for 52 macroevolutionary studies of this pivotal clade. 53