Percutaneous endoscopic lumbar discectomy (PELD) has become an increasingly popular minimally invasive spinal surgery. Due to the technical evolution of PELD, the focus of decompression has shifted from the central nucleus to the subannular-protruded disc herniation, which allows direct neural decompression. Surgical entry into the spinal canal leads to the greater possibility of bony structure obstruction, thus the location and direction of the working channel are crucial. The existing preoperative measuring methods mainly rely on 2-dimensional (2D) x-ray images or MRI cross-sections. Because the bony structure and the trajectory are 3-dimensional (3D), the relationship between the anatomical lumbar structure and the working channel cannot be precisely evaluated. To investigate a 3D method and quantitatively evaluate the trajectory for percutaneous endoscopic lumbar discectomy (PELD). Technical note. Pain medicine center of a university hospital. Twenty patients suffering from L4/5 disc herniation were enrolled in this study. After reconstructing the preoperative CT images, the virtual trajectory was placed into the intervertebral foramen through gradient-changing angulations in relation to the coronal and transverse planes. The overlapping portion of the virtual trajectory and the lumbar vertebrae was evaluated. In addition, the probability of atypical structure involvement was calculated. As cephalad angulation (CA) increased, the intersection volume of the L4 inferior articular process increased, while the total intersection volume, the intersection volume of the L5 superior articular process, the intersection volume of the facet joint, and the volume proportion of L5 superior articular process intersection in the facet joint all decreased. As coronal plane angulation (CPA) increased, the total intersection volume, the intersection volume of the L4 inferior articular process, and the intersection volume of the facet joint all increased, while the volume proportion of the L5 superior articular process intersection in the facet joint decreased. When CA increased to 15°-20°, there was a high probability of atypical structure involvement, whereas such a probability in the groups of CA 0° (CPA 15°, 20°, and 25°), CA 5° and CA 10° was low. Only patients with L4/5 herniation were evaluated in this study. In terms of the regularity, the ideal angulation for L4/L5 PELD is CPA 5°-10° and CA 5°-10°, which can lead to a relatively low level of total damage to the bony structure, minimal damage to the facet joint, and negligible involvement of atypical structures. Lumbar disc herniation, percutaneous endoscopic lumbar discectomy (PELD), transforaminal, trajectory, 3D method, quantitative measurement, angulation, bony structure obstruction.