Spinal cord arteriovenous shunts (SCAVSs) are either fistulas or niduses that can be separated in four different groups according to their localization and relationship to the dura. Paraspinal AVSs are located outside the spine and are responsible for neurological symptoms because of cord compression by ertatic veins, venous congestion or arterial steal. Epidural shunts are located in the epidural space and drain in epidural veins with secondary intradural congestion. Dural shunts are embedded in the dura, produce a cord venous myelopathy after draining through veins that either pierce the dura far from a nerve root or accompany a nerve root. Intradural shunts affect the cord, the roots or the filum. Additionally, they can be classified according to their potential relationships with genetics, vascular biological features and angiogenesis into genetic hereditary lesions (hereditary hemorrhagic telangiectasia), genetic non-hereditary lesions (multiple lesions with metameric links) and single lesions (AVMs or micro AVFs). MRI and MRA are able to visualise SCAVS early after the onset of clinical symptoms. The type of shunt and its localization may remain difficult to be precise. Angiography remains the gold standard for analysis of the anatomical, morphological and architectural features necessary for therapeutic decisions in both paediatric and adult populations. In our series, embolisation is chosen in first intention whatever the type of shunt responsible for the clinical symptoms and glue is preferably used. In paraspinal, dural or epidural arteriovenous shunts, the goal of treatment should be complete closure of the shunt. A complete cure by embolization is rather easily achieved in paraspinal lesions. Failure of endovascular therapy in dural or epidural shunts must bring the patient to surgery. The prognosis of most intradural shunts seems better than previously thought, even after haemorrhage. In intradural spinal cord arteriovenous shunts, embolisation targeted towards the portions of the malformation felt to be responsible for the symptoms (venous congestion) or pointing to the point of rupture (false aneurysms) of the malformation, allows restoration of a new hemodynamic equilibrium between the malformation and the cord itself. Such targeted treatment offers long-term stabilisation or improvement to patients suffering from SCAVSs and good protection against (re) haemorrhages, with an acceptable morbidity. Cure of the shunt is not imperative to obtain these satisfactory outcomes. The clinical results obtained by such management compare favourably with those obtained by neurosurgery.