Abstract Background Despite rapid developments in devices used to treat arteriovenous malformation (AVM), a randomised trial of Unruptured Brain Arteriovenous malformations published in 2014 recommended conservative treatment for nonhemorrhagic AVM. The purpose of the current retrospective study was to confirm how AVM treatment in Japan has changed and to assess the safety of treatment for hemorrhagic and nonhemorrhagic AVMs. Methods We enrolled 242 consecutive patients with AVM; each patient’s treatment was selected and performed at our hospital. The type of onset, Spetzler–Martin (S–M) grade, age, sex, selected treatment, mortality, and morbidity were compared between the first and second periods of our study. Results In patients with grade I–III AVM, the selected treatment changed between the first and second periods; however, in grade IV and V patients, the selected treatment did not change. Overall, interventions by microsurgery alone decreased (p < 0.001), the proportion of total treatments including microsurgery decreased (p = 0.005), interventions using stereotactic radiosurgery (SRS) alone increased (p = 0.009), and interventions including SRS increased (p = 0.002). Morbidity associated with intervention was 0.92% in the first period and 0% in the second period, and mortality was 0.92% in the first period and 1.67% in the second. Conclusions With the development of new devices, the selected treatment was changed in patients with S–M grade I–III AVM, but was not changed in patients with grade IV and V. The complication rate was low and did not change throughout the periods. These findings suggest that the safety of treatment depends on a full understanding of device development and the selection of proper treatment, not on hemorrhagic onset. Further treatment innovations are expected to change the treatment for grade IV and V AVMs.