The incidence for hemorrhage in non-obliterated arteriovenous malformations (AVM) during the first two years following Gamma Knife (GK) surgery was compared to the calculated incidence in untreated patients. There was a decrease in the incidence of hemorrhage as compared to the natural course. This difference was statistically significant. The actual number of hemorrhages during the first two years was 49 in a series of 1604 patients. Additionally, 41 hemorrhages occurred beyond the two years following treatment over the entire follow up time span of up to 24 years. For the first two years the calculated incidence was roughly two times higher than the observed incidence. This impact on the natural history was detectable already within six months after the treatment for the malformations totally covered with at least 25 Gy. The relation between a high minimum and a high average dose on the one hand and a lower risk for hemorrhage on the other hand was statistically significant. Low treatment doses and high age correlated to a higher risk. Neither hemorrhage prior to treatment nor time interval between presenting hemorrhage and treatment seemed to influence the risk for post treatment hemorrhage. The risk for permanent neurological deficit or death due to AVM rupture during the latency period between the time of treatment and total nidus obliteration was less than 0.5% for small AVM and 2-4% for large ones during the first two years. Of the 24 patients with sequelae following AVM rupture after treatment 14 died and 10 remained with neurological deficit of different degrees. However, all survivors were self sufficient.