Atherosclerotic plaques contain a large monoclonal population of cells. Monoclonality could arise by somatic mutation, selection of a pre-existing lineage, or expansion of a pre-existing (developmental) clone. To determine the monoclonal cell type in plaque and learn when monoclonality arises, we studied X chromosome inactivation patterns using methylation of the X-linked human androgen receptor gene. Assays based on polymerase chain reaction were performed on samples of known cellular composition, microdissected from histological sections of human arteries. In atherosclerotic vessels, the majority of medial samples (7/11 coronary and 2/3 aortic) showed balanced (paternal and maternal) patterns of X inactivation, indicating polyclonality. In contrast, most samples of plaque smooth muscle cells showed a single pattern of X inactivation (3/4 aortic plaques and 9/11 coronary plaques; P< 0.01 versus media), indicating that plaque smooth muscle cells are monoclonal. Samples of plaque containing inflammatory or endothelial cells showed balanced X inactivation, also demonstrating polyclonality. Multiple plaques from a given patient showed no bias toward one allele, indicating there was no X-linked selection of cells during plaque growth. To determine whether plaques might arise from pre-existing clones (large X inactivation patches), we then studied 10 normal coronaries with diffuse intimal thickening. Six of the ten coronaries showed skewed X inactivation patterns in normal media and intima, suggesting the patch size in normal arteries is surprisingly large. Thus, smooth muscle cells constitute the monoclonal population in atherosclerotic plaques. The finding that normal arteries may have large X inactivation patches raises the possibility that plaque monoclonality may arise by expanding a pre-existing clone of cells rather than generating a new clone by mutation or selection.