Object The aim of this study was to identify patients who are likely to benefit from surgery for unruptured brain arteriovenous malformations (ubAVMs). Methods The authors' database was interrogated for the risk and outcome of hemorrhage after referral and the outcome from surgery. Furthermore, the outcome from surgery incorporated those cases excluded from surgery because of perceived greater risk (sensitivity analysis). Finally, a comparison was made for the authors' patients between the natural history and surgery. Data were collected for 427 consecutively enrolled patients with ubAVMs in a database that included patients who were conservatively managed. Kaplan-Meier analysis was performed on patients observed for more than 1 day to determine the risk of hemorrhage. Variables that may influence the risk of first hemorrhage were assessed using Cox proportional hazard regression models and Kaplan-Meier life table analyses from referral until the first occurrence of the following: hemorrhage, treatment, or last review. The outcome from surgery (leading to a new permanent neurological deficit with last review modified Rankin Scale [mRS] score > 1) was determined. Further sensitivity analysis was made to predict risk from surgery for the total ubAVM cohort by incorporating outcomes of surgical cases as well as cases excluded from surgery because of perceived risk, and assuming an adverse outcome for these excluded cases. Results A total of 377 patients with a ubAVM were included in the analysis of the risk of hemorrhage. The 5-year risk of hemorrhage for ubAVM was 11.5%. Hemorrhage resulted in an mRS score > 1 in 14 cases (88% [95% CI 63%-98%]). Patients with Spetzler-Ponce Class A ubAVMs treated by surgery (n = 190) had a risk from surgery of 1.6% (95% CI 0.3%-4.8%) for a permanent neurological deficit leading to an mRS score > 1 and 0.5% (95% CI < 0.1%-3.2%) for a permanent neurological deficit leading to an mRS score > 2. Patients with Spetzler-Ponce Class B ubAVMs treated by surgery (n = 107) had a risk from surgery of 14.0% (95% CI 8.6%-22.0%) for a permanent neurological deficit leading to an mRS score > 1. Sensitivity analysis of Spetzler-Ponce Class B ubAVMs, including those in patients excluded from surgery, showed that the true risk for surgically eligible patients may have been as high as 15.6% (95% CI 9.9%-23.7%) for mRS score > 1, had all patients who were perceived to have a greater risk experienced an adverse outcome. Patients with Spetzler-Ponce Class C ubAVMs treated by surgery (n = 44) had a risk from surgery of 38.6% (95% CI 25.7%-53.4%) for a permanent neurological deficit leading to an mRS score > 1. Sensitivity analysis of Class C ubAVMs, including those harbored by patients excluded from surgery, showed that the true risk for surgically eligible patients may have been as high as 60.9% (95% CI 49.2%-71.5%) for mRS score > 1, had all patients who were perceived to have a greater risk experienced an adverse outcome. Conclusions Surgical outcomes for Spetzler-Ponce Class A ubAVMs are better than those for conservative management.