OBJECTIVE Grade II meningiomas, which currently account for 25% of all meningiomas, are subject to multiple recurrences throughout the course of the disease and represent a challenge for the neurosurgeon. Radiosurgery is increasingly performed for the treatment of Grade II meningiomas and is quite efficient in controlling relapses locally at the site of the lesion, but it cannot prevent margin relapses. The aim of this retrospective study was to analyze the technical parameters involved in producing marginal relapses and to optimize loco-marginal control to improve therapeutic strategy. METHODS Eighteen patients presenting 58 lesions were treated by Gamma Knife radiosurgery (GKRS) between 2010 and 2015 in Hopital de la Pitié-Salpêtrière. The median patient age was 68 years (25%-75% interval: 61-72 years), and the sex ratio (M/F) was 13:5. The median delay between surgery and first GKRS was 3 years. Patients were classified as having Grade II meningioma using World Health Organization (WHO) 2007 criteria. The tumor growth rate was computed by comparing 2 volumetric measurements before treatment. After GKRS, iterative MRI, performed every 6 months, detected a relapse if tumor volume increased by more than 20%. Patterns of relapse were defined as being local, marginal, or distal. Survival curves were estimated using the Kaplan-Meier method, and the relationship between criterion and potential risk factors was tested by the log-rank test and univariable Cox model. RESULTS The median follow-up was 36 months (range 8-57 months). During this period, 3 patients presented with a local relapse, 5 patients with a marginal relapse, and 7 patients with a distal relapse. Crude local control was 84.5%. The local control actuarial rate was 89% at 1 year and 71% at 3 years. The marginal control actuarial rate was 81% at 1 year and 74% at 2 years. The distal control actuarial rate was 100% at 1 year, 81% at 2 years, and 53% at 3 years. Median distal control was 38 months. Progression-free survival (PFS) was 71% at 1 year, 36% at 2 years, and 23% at 3 years. Median PFS was 18 months. Lesions treated with a minimum radiation dose of ≤ 12 Gy had significantly more local relapses than those treated with a dose > 12 Gy (p = 0.04) in univariate analysis. Marginal control was significantly influenced by tumor growth rate, with a lower growth rate being highly associated with improved marginal control (p = 0.002). There was a trend toward a relationship between dose and marginal control, but it was not significant (p = 0.09). PFS was significantly associated with delay between first surgery and GKRS (p = 0.03). The authors noticed few complications with no sequelae. CONCLUSIONS In order to optimize loco-marginal control, radiosurgical treatment should require a minimum dose of > 12 Gy and an extended target volume along the dural insertion. Ideally, these parameters should correspond to the aggressiveness of the lesion, based on genetic features of the tumor.