Objective To investigate the application and safety of stereoelectroencephalography (SEEG) in invasive preoperation evaluation of drug-refractory epilepsy (DRE). Methods We retrospectively analyzed the clinical data of patients with DRE who underwent SEEG implantation in Shenzhen Second People's Hospital and Shenzhen University General Hospital between Aug. 2016. and Nov. 2018. The suspicious epileptogenic focus and propagation path was proposed based on the non-invasive preoperative evaluation, and then the implantation protocol of the SEEG electrodes was designed. Finally, according to the SEEG and electrical stimulation results, the resection or thermocoagulation of suspicious epileptogenic focus was done. Results A total of 359 electrodes were implanted in 31 DRE patients, with an average of 11.58 per patient. There were 13 electrodes (3.62%) failed to be implanted, with no additional electrode implantation. One case of intracranial hemorrhage was noted after operation, and there was no intracranial infection or cerebrospinal fluid leakage. Twenty patients received epileptogenic zone resection, 8 received epileptic network radiofrequency thermocoagulation, 1 with epileptic foci in the eloquent cortex underwent vagus nerve stimulation, and 2 received no operation. Thirty-one DRE patients were followed up for 6-30 months, with an average of (11.68±7.46) months. Of the 20 patients who underwent epileptogenic zone resection, 17 patients (85.00%, 17/20) had grade Ⅰ ILAE (International League Against Epilepsy), 2 (10.00%, 2/20) had grade Ⅱ, and 1 (5.00%, 1/20) had grade Ⅲ. For the 8 patients undergoing epileptic network radiofrequency thermocoagulation, 6 (75.00%, 6/8) had grade ⅠILAE and 2 (25.00%, 2/8) had grade Ⅲ. There was no significant difference in outcome between the two treatments (P=0.61). For 13 patients with negative magnetic resonance imaging (MRI), 10 (76.92%, 10/13) had grade Ⅰ ILAE, 2 (15.38%, 2/13) had grade Ⅱ, and 1 (7.69%, 1/13) had grade Ⅲ; for 15 MRI-positive patients, 13 (86.67%, 13/15) had grade Ⅰ ILAE and 2 (13.33%, 2/15) had grade Ⅲ. There was no significant difference in outcome between MRI-positive patients and MRI-negative patients (P=0.64). Conclusion Surgical treatment of DRE patient under the guidance of SEEG is safe and effective, and SEEG can improve the accuracy of epileptogenic foci localization, no matter in MRI positive or MRI negative patients.