PURPOSE: To assess the embolization effect of dural arteriovenous fistula(DAVF). MATERIALS AND METHODS: Weevaluated 23 patients with DAVF who were diagnosed using digital subtraction angiography and treated bytransarterial and/or transvenous emobolization. The locations of DAVFs, as seen on angiography were as follows: 16in the cavernous sinus, four in the major dural sinuses, two in the jugular bulb, one in the torcula Herophili,and one in the tentorium cerebelli. On the basis of venous drainage patterns, Cognard's classification of DAVF wasused. Among our 23 patients, 13 underwent transarterial embolization, six underwent transvenous embolization, andfour underwent both. After embolization, three patients underwent other treatment: surgery, one ; gammairradiation, one ; and both. One. Polyvinyl alcohol particles or glue was used in transarterial embolization, andtungsten, platinum, or Guglielmi detachable coils in transvenous embolization. To analyse the residual lesion inDAF, we compared pre- and post-treatment angiograms ; patients were followed up for 2 to 48 months, and theircondition was assessed as cured, improved, not changed, or aggravated. RESULTS: Nine patients were found to beCognard type I, two were IIa, four were IIa+b, and eight were type III. Of our 23 patients, 12 were cured, tenimproved, and one was aggravated ; of the 13 who underwent transarterial embolization, four were cured, eightimproved, and one was aggravated ; of the six who underwent transvenous embolization, five were cured and oneimproved. The condition of a patient with a lesion in the torcular herophili was aggravated despite surgery andgamma irradiation after embolization. CONCLUSION: DAVF can be managed successfully with endovascular treatment.The outcome of the transvenous approach appears to be better than that of the transarterial approach alone. In acase involving a lesion in the torcular herophili, the outcome was poor and more aggressive treatment wasrequired.