Limbal stem cell deficiency (LSCD) is a common ocular surface disease.Severe LSCD can cause irreversible vision loss and the only effective treatment for severe LSCD is keratoplasty, which is limited by the number of donors.Exploring new effective treatments in lieu of keratoplasty remains a hot topic in current research.Since new treatments need to be critically evaluated in reliable animal models that reflect the pathological mechanisms of human LSCD, the establishment of an appropriate animal model is crucial in any experiment.Up to now, the main modeling methods for LSCD are mechanical method, thermal injury, chemical method, chemical method combined with surgery, and gene knock-down.Mechanical method is mature, reliable and widely used, but it can cause serious mechanical damage.Rotate burr tool could be the best mechanical method because of uniform damage and the small trauma, but it requires accumulation of experience.Cornea suffered from thermal injury performs stable, but the operation is complicated and uncontrollable.Chemical method is quick and simple, and the etiology is highly consistent with clinical cases, but the depth and extent of injury are uncontrollable.The chemical method combined with surgery could ensure the smoothness and flattening of the corneal surface, but the thickness of the central cornea and the limbus may be inconsistent, which affects neovascularization.The gene knockdown method is not widely used because of its complicated operations.Although there are different types of methods, the quality of modeling in different studies is not consistent.Therefore, selecting a reliable and efficient animal model of LSCD is not convenient for researchers.This review discussed current methods to establish LSCD model, including their advantages and disadvantages, therefore to provide inspiration for future model improvement.