A major concern of patients who have stereotactic radiosurgery is the long-term risk of having a secondary intracranial malignancy or, in the case of patients with benign tumours treated with the technique, the risk of malignant transformation. The incidence of stereotactic radiosurgery-associated intracranial malignancy remains unknown; therefore, our aim was to estimate it in a population-based study to assess the long-term safety of this technique. We did a population-based, multicentre, cohort study at five international radiosurgery centres (Na Homolce Hospital, Prague, Czech Republic [n=2655 patients]; Ruber International Hospital, Madrid, Spain [n=1080], University of Pittsburgh Medical Center, Pittsburgh, PA, USA [n=1027]; University of Virginia, Charlottesville, VA, USA [n=80]; and NYU Langone Health System, New York, NY, USA [n=63]). Eligible patients were of any age, and had Gamma Knife radiosurgery for arteriovenous malformation, trigeminal neuralgia, or benign intracranial tumours, which included vestibular or other benign schwannomas, WHO grade 1 meningiomas, pituitary adenomas, and haemangioblastoma. Patients were excluded if they had previously had radiotherapy or did not have a minimum follow-up time of 5 years. The primary objective of the study was to estimate the incidence of stereotactic radiosurgery-associated intracranial malignancy, including malignant transformation of a benign lesion or development of radiation-associated secondary intracranial cancer, defined as within the 2 Gy isodose line. Estimates of age-adjusted incidence of primary CNS malignancies in the USA and European countries were retrieved from the Central Brain Tumor Registry of the United States (CBTRUS) and the International Agency for Research on Cancer (IARC) Global Cancer statistics. Of 14 168 patients who had Gamma Knife stereotactic radiosurgery between Aug 14, 1987, and Dec 31, 2011, in the five contributing centres, 4905 patients were eligible for the analysis (had a minimum follow-up of 5 years and no history of previous radiation therapy). Diagnostic entities included vestibular schwannomas (1011 [20·6%] of 4905 patients), meningiomas (1490 [30·4%]), arteriovenous malformations (1089 [22·2%]), trigeminal neuralgia (565 [11·5%]), pituitary adenomas (641 [13·1%]), haemangioblastoma (29 [0·6%]), and other schwannomas (80 [1·6%]). With a median follow-up of 8·1 years (IQR 6·0-10·6), two (0·0006%) of 3251 patients with benign tumours were diagnosed with suspected malignant transformation and one (0·0002%) of 4905 patients was considered a case of radiosurgery-associated intracranial malignancy, resulting in an incidence of 6·87 per 100 000 patient-years (95% CI 1·15-22·71) for malignant transformation and 2·26 per 100 000 patient-years (0·11-11·17) for radiosurgery-associated intracranial malignancy. Two (0·0004%) of 4905 patients developed intracranial malignancies, which were judged unrelated to the radiation field. Overall incidence of radiosurgery-associated malignancy was 6·80 per 100 000 patients-years (95% CI 1·73-18·50), or a cumulative incidence of 0·00045% over 10 years (95% CI 0·00-0·0034). The overall incidence of 6·8 per 100 000, which includes institutions from Europe and the USA, after stereotactic radiosurgery was found to be similar to the risk of developing a malignant CNS tumour in the general population of the USA and some European countries as estimated by the CBTRUS and IARC data, respectively. These data show that the estimated risk of an intracranial secondary malignancy or malignant transformation of a benign tumour in patients treated with stereotactic radiosurgery remains low at long-term follow-up, and is similar to the risk of the general population to have a primary CNS tumour. Although prospective cohort studies with longer follow-up are warranted to support the results of this study, the available evidence suggests the long-term safety of stereotactic radiosurgery and could support physicians counselling patients on Gamma Knife stereotactic radiosurgery. None.