The assessment of various radiation-induced effects on important white matter tracts including the limbic circuit fibers is helpful to reduce toxicity. According to long-term follow up of EORTC 26951 trials, adding 6 cycles of procarbazine, lomustine, and vincristine (PCV) chemotherapy to radiotherapy (RT) prolonged survival in anaplastic oligodendroglioma (AO) patients. Consequently, this raised the interest the radiation induced long-term toxicity. We aimed to analyze the relation between RT dose of fornix and symptomatic radiation necrosis (SRN) in patients with AO treated with surgery and RT in this preliminary study. A 67 AO patients treated between Feb. 2009 and Jul. 2019 were analyzed. SRN was defined when three criteria were met: (1) radiographic findings were confirmed, (2) the patient experienced new or progressive symptoms that were attributable to the progressing lesion, and (3) treatment with corticosteroids, bevacizumab, or surgical resection only resulted in improvement of symptoms without further radiographic progression of the lesion. Various contours, including fornix, corpus callosum, basal ganglia, and septum pellucidum, are delineated manually on simulation CT fused with preoperative and postoperative MRI. Patients received 3 D-CRT (n = 17, 25%) or IMRT (n = 50, 75%) with median RT dose of 60 Gy (range, 52.5-70). Univariate logistic regression models were fitted to examine whether maximum and mean radiation dose of contours and treatment variables predicted the development of SRN. Receiver operating characteristic curve (ROC) analysis was used to assess the cut-point of radiation dose predicting SRN. With median follow-up of 42 months (range 9-125) for surviving patients, median overall survival and progression survival were 74 and 74 months, respectively. Median age was 41 years (range, 17-78) and median KPS was 90 (range, 50-100). The most common tumor locations were frontal (n = 49, 60%) and parietal (n = 16, 24%). Gross total resection was done for 38 patients (57%) and subtotal resection was done for 23 patients (34%). There were 42 patients (63%) received PCV chemotherapy. There were 17 patients (25%) who had SRN. Univariate analysis revealed that maximum dose of fornix (p = 0.011), mean dose of anterior (p = 0.033) and posterior fornix (p = 0.021) predicted the development of SRN. The maximum dose of fornix (equivalent total dose in 2 Gy fraction) was found to have the largest areas under the ROC curve (0.755, 95% CI 0.628-0.883), and the cut-off values for SRN was 59 Gy (EQD2). The rate of development of SRN was significantly higher in patients whose maximum dose of fornix more than 59 Gy (13 vs. 43%, p = 0.005). In our study, maximum dose of fornix is significant factor for development of SRN and cut point is 59 Gy (EQD2). Fornix sparing using modern RT technique is helpful to maintain neurocognitive function, and additional study is needed to evaluate the difference of recurrence and survival regarding fornix sparing.