BACKGROUND: The anatomy of carotid stenosis may influence the outcome of endovascular treatment or carotid endarterectomy. Whether anatomy favors one treatment over the other in terms of safety or efficacy has not been investigated in randomized trials. METHODS: In 414 patients with mostly symptomatic carotid stenosis randomized to endovascular treatment (angioplasty or stenting; n = 213) or carotid endarterectomy (n = 211) in the Carotid and Vertebral Artery Transluminal Angioplasty Study (CAVATAS), the degree and length of stenosis and plaque surface irregularity were assessed on baseline intraarterial angiography. Outcome measures were stroke or death occurring between randomization and 30 days after treatment, and ipsilateral stroke and restenosis ≥50% during follow-up. RESULTS: Carotid stenosis longer than 0.65 times the common carotid artery diameter was associated with increased risk of peri-procedural stroke or death after both endovascular treatment [odds ratio 2.79 (1.17-6.65), P = 0.02] and carotid endarterectomy [2.43 (1.03-5.73), P = 0.04], and with increased long-term risk of restenosis in endovascular treatment [hazard ratio 1.68 (1.12-2.53), P = 0.01]. The excess in restenosis after endovascular treatment compared with carotid endarterectomy was significantly greater in patients with long stenosis than with short stenosis at baseline (interaction P = 0.003). Results remained significant after multivariate adjustment. No associations were found for degree of stenosis and plaque surface. CONCLUSIONS: Increasing stenosis length is an independent risk factor for peri-procedural stroke or death in endovascular treatment and carotid endarterectomy, without favoring one treatment over the other. However, the excess restenosis rate after endovascular treatment compared with carotid endarterectomy increases with longer stenosis at baseline. Stenosis length merits further investigation in carotid revascularisation trials.