Objectives-To investigate the contribution of plaque size and vessel remodelling to coronary artery stenosis and to assess the role of vessel shrinkage (negative remodelling) across a wide range of lesions. Design-Postmortem study of coronary remodelling in perfusion fixed hearts. Subjects-24 men and 24 women who died suddenly with coronary artery disease. Main outcome measures-Percentage stenosis, percentage plaque burden, percentage remodelling, and arc of normal vessel were measured and related to age, sex, smoking status, and history of hypertension. Results-There was a positive relation between percentage stenosis and percentage plaque burden (r = 0.6, p < 0.0001) and an inverse relation between percentage stenosis and percentage remodelling (r = -0.4, p < 0.0001). Multilinear regression modelling showed that luminal stenosis = 1.0 (plaque burden) - 0.4 (vessel remodelling). Remodelling was greater in lesions that would not have been significant at angiography (less than or equal to 25% stenosis) than in the remaining lesions (25.9 (26)% v 10.0 (21.1)%, p < 0.0001, respectively) and was reduced in segments with circumferential plaques (12.7 (24.5) % v 20.7 (24.3) % in eccentric plaques, p = 0.001). Remodelling did not correlate with age, sex, or smoking. Negative remodelling was present in 62 lesions with a stenosis > 25% versus 10 lesions with less than or equal to 25% stenosis (p < 0.0001). Lesions with negative remodelling had greater plaque burden and luminal stenosis and a reduced arc of normal segment. Conclusion-Outward arterial remodelling negates the stenosing effect of increasing plaque size. Significant coronary stenoses arise from a failure of this outward remodelling in the face of a large plaque burden. Coronary arterial remodelling is unrelated to sex or smoking and is plaque specific.