Only a small proportion of genetic variation in complex traits has been explained by SNPs from genome-wide association studies (GWASs). We report the results from two GWASs for serum markers of iron status (serum iron, serum transferrin, transferrin saturation with iron, and serum ferritin), which are important in iron overload (e.g., hemochromatosis) and deficiency (e.g., anemia) conditions. We performed two GWASs on samples of Australians of European descent. In the first GWAS, 411 adolescent twins and their siblings were genotyped with 100K SNPs. rs1830084, 10.8 kb 3' of TF, was significantly associated with serum transferrin (p total association test = 1.0 x 10(-9); p within-family test = 2.2 x 10(-5)). In the second GWAS on an independent sample of 459 female monozygotic (MZ) twin pairs genotyped with 300K SNPs, we found rs3811647 (within intron 11 of TF, HapMap CEU r(2) with rs1830084 = 0.86) was significantly associated with serum transferrin (p = 3.0 x 10(-15)). In the second GWAS, we found two additional and independent SNPs on TF (rs1799852 and rs2280673) and confirmed the known C282Y mutation in HFE to be independently associated with serum transferrin. The three variants in TF (rs3811647, rs1799852 and rs2280673) plus the HFE C282Y mutation explained approximately 40% of genetic variation in serum transferrin (p = 7.8 x 10(-25)). These findings are potentially important for our understanding of iron metabolism and of regulation of hepatic protein secretion, and also strongly support the hypothesis that the genetic architecture of some endophenotypes may be simpler than that of disease.