Purpose To develop a 3D anatomical model for teaching canine epidural anesthesia (3DMEA) and to assess its efficacy for teaching and learning prior to the use of live animals. Methods The creation of 3DMEA was based on 3D optical scanning and 3D printing of canine bone pieces of the fifth to the seventh lumbar vertebrae, sacrum and pelvis. A total of 20 male dogs were scheduled for castration. 20 veterinary students watched a video showing epidural anesthesia in dogs before the clinical attempt and were assigned to control or 3DMEA groups. Students in the 3DMEA group trained in the model after the video. For the clinical trial, the epidural procedure was performed by students under the veterinary supervision. When observed the absence of response to nociceptive stimuli, the epidural was considered successful. Then, all students answered a questionnaire evaluating the main difficulty founded in the technique and its degree of difficulty. Results The 3DMEA group reported a lower degree of difficulty to perform the epidural anesthesia technique when compared with the control group (p=0.0037). The 3DMEA reproduced the anatomical structures, allowing the perception of the distance of needle in relation to the iliac prominences during epidural anesthesia. Its mobility allowed simulation of the animal in standing position and sternal recumbency. Conclusion The use of 3DMEA demonstrated greater efficacy in the execution of the technique, being effective in the teaching and learning process before the epidural anesthesia in live animals.