The median overall survival of glioblastoma patients is about 15 months. Only a small number of patients survive 3 years. The factors of a favorable prognosis for the «longevity phenomenon» in glioblastoma patients are not fully understood. Objective — to determine the occurrence rate of long-living patients with glioblastomas, identify clinical predictors of a favorable prognosis, and identify the presence and prognostic significance of an IDH1 mutation. Material and methods. Among 1494 patients operated on for glioblastoma at the Burdenko Neurosurgical Institute from 2007 to 2012, there were 84 (5.6%) patients who lived more than 3 years after primary surgery. In all the cases, histological specimens were reviewed, and immunohistochemical detection of a mutant IDH1 protein was performed. Overall survival was calculated from the time of first surgery to the date of the last consultation or death, and the recurrence-free period was calculated from the time of first surgery to MRI-verified tumor progression. Results. The median age of long-living patients with glioblastoma was 45 years (19―65 years). All tumors were located supratentorially. The median Karnofsky performance status score at the time of surgery was 80 (range, 70―100). All patients underwent microsurgical resection of the tumor, followed by chemoradiotherapy. The median recurrence-free period was 36 months (5―98 months). Overall survival of 48, 60, and 84 months was achieved in 23, 15 and 6% of patients, respectively. Among 49 specimens available for the IDH1 analysis, 14 (28.6%) specimens had a mutant protein. There was no significant difference in survival rates in patients with positive and negative results for IDH1 (44.1 vs. 40.8 months; p>0.05). Conclusion. The significance of various factors that may be predictors of a favorable course of the disease is discussed in the literature. This work is the first part of analysis of prognostically significant factors positively affecting overall survival of glioblastoma patients. In our series, the predictors of a favorable prognosis for long-living patients with the verified diagnosis of glioblastoma were as follows: young age, the supratentorial location of the tumor, a high Karnofsky score before surgery, and tumor resection. In our series, we used immunohistochemical tests and found no prognostic significance of the IDH1 gene mutation; further analysis will require application of direct sequencing. We plan to study other morphological and molecular genetic features of tumors, which explain prolonged survival of glioblastoma patients, as well as the role of various types of combined chemoradiation treatment.