Spinal manipulation is a form of manual therapy often employed by chiropractors, osteopaths, medical doctors, physiotherapists and other healthcare professionals. Its widespread use renders the continuous, critical assessment of its safety an important task. The aim of this systematic review was to summarise the recent (1995-2003) evidence on the subject. Computerised literature searches were carried out without language restrictions. Only articles with original data pertaining to the cerebrovascular risks of spinal manipulation were included. Numerous case reports, case series, retrospective investigations and prospective studies were located. In total, more than 300 patients were affected. The most frequently reported complication was stroke due to arterial dissection after cervical spinal manipulation. Considering the popularity of spinal manipulative therapies, the overall incidence of such complications is probably low; however, no reliable figures can be generated through this or any other data available to date. It is concluded that serious cerebrovascular complications of spinal manipulation continue to be reported. Their incidence is unknown. Large and rigorous prospective studies are necessary in order to define the risks of spinal manipulation accurately.