Limbal stem cell defect model is an important animal model that provides a basis for the study of ocular surface diseases. The rabbit cornea is of moderate size and is widely used in such studies as an experimental animal model. At present, the main modeling methods are alkali burns, and corneal limbus girdling and corneal epithelium doctoring. Each method has its own characteristics. In this study, we observed rabbit models with severe ocular surface defect established by the two methods and changes after amniotic membrane transplantation. In the first, second, third, and fourth week after operation, the clinical manifestations, corneal transparency, and new vessels were observed according to the standard rating scale of ocular surface, compared between the two methods, and then statistically analyzed. In the fourth week after operation, the rabbits were sacrificed and their corneas and corneal limbus were extracted from sclera, embedded by optimum cutting temperature compound, frozen, and sliced for hematoxylin and eosin staining and pathological examination. There were two groups in this study. Group 1 (alkali burns) had more severe complications, such as, conjunctiva, nubecula, new vessel hyperplasia, and so on, compared to group 2 (corneal limbus girdling and corneal epithelium doctoring). In addition, there were striking differences in corneal transparency and new vessels between the two groups (p<0.05). Corneal transparency in group 1 was lower than in group 2. New vessels in group 1 were less in the first 2 weeks, but obviously increased compared to group 2 in the subsequent weeks. Alkaline burn could be used to study new vessel hyperplasia, while corneal limbus girdling and corneal epithelium doctoring are more suitable for studying stem cell transdifferentiation, interactive roles of stem cells and microenvironment, and so on.