An augmented reality (AR)-assisted surgical navigation system was developed for epidural needle intervention. The system includes three components: a virtual reality-based surgical planning software, a patient and tool tracking system, and an AR-based surgical navigation system. A three-dimensional (3D) path plan for the epidural needle was established on the preoperative computed tomography (CT) image. The plan is then registered to the intraoperative space by 3D models of the target vertebrae using skin markers and real-time tracking information. In the procedure, the plan and tracking information are transmitted to the head-mounted display (HMD) through a wireless network such that the device directly visualizes the plan onto the back surface of the patient. The physician determines the entry point and inserts the needle into the target based on the direct visual guidance of the system. An experiment was conducted to validate the system using two torso phantoms that mimic human respiration. The experimental results demonstrated that the time and the number of X-rays required for needle insertion were significantly decreased by the proposed method (43.6±20.55sec, 2.9±1.3times) compared to those of the conventional fluoroscopy-guided approach (124.5 ± 46.7s, 9.3±2.4times), whereas the average targeting errors were similar in both cases. The proposed system may potentially decrease ionizing radiation exposure not only to the patient but also to the medical team.