Long-term solid-organ allografts typically develop diffuse arterial intimal lesions (graft arterial disease; GAD), consisting of smooth-muscle cells (SMC), extracellular matrix and admixed mononuclear leukocytes. GAD eventually culminates in vascular stenosis and ischemic graft failure. Although the exact mechanisms are unknown, chronic low-level alloresponses likely induce inflammatory cells and/or dysfunctional vascular wall cells to secrete growth factors that promote SMC intimal recruitment, proliferation and matrix synthesis 1, 2, 3. Although prior work demonstrated that the endothelium and medial SMCs lining GAD lesions in cardiac allografts are donor-derived, the intimal SMC origin could not be determined 4. They are generally presumed to originate from the donor media 5, leading to interventions that target donor medial SMC proliferation, with limited efficacy 6, 7. However, other reports indicate that allograft vessels may contain host-derived endothelium and SMCs (refs. 8, 9). Moreover, subpopulations of bone-marrow and circulating cells can differentiate into endothelium 10, 11, and implanted synthetic vascular grafts are seeded by host SMCs and endothelium 12, 13. Here we used murine aortic transplants to formally identify the source of SMCs in GAD lesions. Allografts in β-galactosidase transgenic recipients showed that intimal SMCs derived almost exclusively from host cells. Bone-marrow transplantation of β-galactosidase–expressing cells into aortic allograft recipients demonstrated that intimal cells included those of marrow origin. Thus, smooth-muscle–like cells in GAD lesions can originate from circulating bone-marrow–derived precursors.