BACKGROUND: Extracranial carotid and vertebral artery dissection is an important cause of stroke, especially in young people. In some observational studies it has been associated with a high risk of recurrent stroke. Both antiplatelet drugs and anticoagulant drugs are used to reduce risk of stroke but whether one treatment strategy is more effective than the other is unknown. We compared their efficacy in the Cervical Artery Dissection in Stroke Study (CADISS), with the additional aim of establishing the true risk of recurrent stroke. METHODS: We did this randomised trial at hospitals with specialised stroke or neurology services (39 in the UK and seven in Australia). We included patients with extracranial carotid and vertebral dissection with onset of symptoms within the past 7 days. Patients were randomly assigned (1:1) by an automated telephone randomisation service to receive antiplatelet drugs or anticoagulant drugs (specific treatment decided by the local clinician) for 3 months. Patients and clinicians were not masked to allocation, but investigators assessing endpoints were. The primary endpoint was ipsilateral stroke or death in the intention-to-treat population. The trial was registered with EUDract (2006-002827-18) and ISRN (CTN44555237). FINDINGS: We enrolled 250 participants (118 carotid, 132 vertebral). Mean time to randomisation was 3·65 days (SD 1·91). The major presenting symptoms were stroke or transient ischaemic attack (n=224) and local symptoms (headache, neck pain, or Horner's syndrome; n=26). 126 participants were assigned to antiplatelet treatment versus 124 to anticoagulant treatment. Overall, four (2%) of 250 patients had stroke recurrence (all ipsilateral). Stroke or death occurred in three (2%) of 126 patients versus one (1%) of 124 (odds ratio [OR] 0·335, 95% CI 0·006-4·233; p=0·63). There were no deaths, but one major bleeding (subarachnoid haemorrhage) in the anticoagulant group. Central review of imaging failed to confirm dissection in 52 patients. Preplanned per-protocol analysis excluding these patients showed stroke or death in three (3%) of 101 patients in the antiplatelet group versus one (1%) of 96 patients in the anticoagulant group (OR 0·346, 95% CI 0·006-4·390; p=0·66). INTERPRETATION: We found no difference in efficacy of antiplatelet and anticoagulant drugs at preventing stroke and death in patients with symptomatic carotid and vertebral artery dissection but stroke was rare in both groups, and much rarer than reported in some observational studies. Diagnosis of dissection was not confirmed after review in many cases, suggesting that radiographic criteria are not always correctly applied in routine clinical practice. FUNDING: Stroke Association.