This paper provides an overview of primary/secondary outcome data from 20 randomised controlled trials comparing carotid endarterectomy (CEA) with carotid artery stenting (CAS) in symptomatic and asymptomatic patients, including meta-analyses for peri-operative risks and late ipsilateral stroke. Secondary analyses include (i) risk factors for stroke after CEA/CAS; (ii) the effect of peri-operative stroke or myocardial infarction on long term survival; (iii) non-stroke complications; (iv) the significance of new white matter lesions on late stroke and cognitive impairment; and (v) whether asymptomatic 70%e99% restenoses increase the risk of ipsilateral stroke after CEA and CAS. Objectives: The aim of this review was to carry out primary and secondary analyses of 20 randomised controlled trials (RCTs) comparing carotid endarterectomy (CEA) with carotid artery stenting (CAS). Methods: A systematic review and meta-analysis of data from 20 RCTs (126 publications) was carried out. Results: Compared with CEA, the 30 day death/stroke rate was significantly higher after CAS in seven RCTs involving 3467 asymptomatic patients (odds ratio [OR] 1.64, 95% confidence interval [CI] 1.02e2.64) and in 10 RCTs involving 5797 symptomatic patients (OR 1.71, 95% CI 1.38e2.11). Excluding procedural risks, late ipsilateral stroke was about 4% at 9 years for both CEA and CAS, i.e., CAS was durable. Reducing procedural death/stroke after CAS may be achieved through better case selection, e.g., performing CEA in (i) symptomatic patients aged > 70 years; (ii) interventions within 14 days of symptom onset; and (iii) situations where stroke risk after CAS is predicted to be higher (segmental/remote plaques, plaque length > 13 mm, heavy burden of white matter lesions [WMLs], where two or more stents might be needed). New WMLs were significantly more common after CAS (52% vs. 17%) and were associated with higher rates of late stroke/transient ischaemic attack (23% vs. 9%), but there was no evidence that new WMLs predisposed towards late cognitive impairment. Restenoses were more common after CAS (10%) but did not increase late ipsilateral stroke. Restenoses (70%e99%) after CEA were associated with a small but significant increase in late ipsilateral stroke (OR 3.87, 95% CI 1.96e7.67; p < .001). Conclusions: CAS confers higher rates of 30 day death/stroke than CEA. After 30 days, ipsilateral stroke is virtually identical for CEA and CAS. Key issues to be resolved include the following: (i) Will newer stent technologies and improved cerebral protection allow CAS to be performed < 14 days after symptom onset with risks similar to CEA? (ii) What is the optimal volume of CAS procedures to maintain competency? (iii) How to deliver better risk factor control and best medical treatment? (iv) Is there a role for CEA/CAS in preventing/reversing cognitive impairment?