This study aimed to evaluate the safety and accuracy of the endoscopic transethmoid-sphenoid approach for optic canal decompression. Twelve sides of 6 adult cadaveric heads fixed in formalin were selected to simulate optic canal decompression using the endoscopic transethmoid-sphenoid approach. Furthermore, this approach was used for optic canal decompression in 10 patients (11 eyes) with optic nerve canal injury. Related anatomical structures were observed using a 0-degree endoscope, and the anatomical characteristics as well as the surgical data were collected. The maximum effective widths of the cranial opening, orbital opening, and middle segment of the canal that could be drilled open endoscopically were 7.82±2.63, 8.05±2.77, and 6.92±2.01 mm, respectively. The angle between the line linking the center point of the tubercular recess with the midpoint of the cranial opening of the optic canal and the horizontal coordinate was 17.23±1.34 degrees. At the orbital opening of the optic canal, the ophthalmic artery was located directly inferior to the optic nerve in 2 cases (16.7%) and laterally inferior to the optic nerve in 10 cases (83.3%). Six of the operational eyes were effective while the remaining 5 were ineffective. No postoperative complications such as bleeding, infection, or cerebrospinal fluid leakage were observed during the follow-up period (6–12 mo). In conclusion, optic canal decompression positively impacts the prognosis of partial traumatic optic neuropathy. Furthermore, the endoscopic transethmoid-sphenoid approach for optic canal decompression is a minimally invasive procedure that provides direct access and adequate decompression. This technique is easy to master and suitable for clinical applications.