Based on a personal case of congenital arteriovenous fistula we discuss the problem of the surgical treatment of this disease. Surgical treatment is always indicated in any case of systemic haemodynamic effects. The surgical treatment is not always successful as the fistula cannot always be exactly located. Another problem is the development of new circuits in regions mo-e proximally situated. This results in a high volume shunt, even if the more peripheral anomalies seem to have been removed completely. It is not sufficient to disconnect and resect the main arteries, without removing the fistula or the angioma, or new collateral vessels and trophic changes will be seen.