The magnitude of the deformations around excavations is an important criterion in the design of the support system of an excavation. In this study, the deformation behaviour of Diaphragm Wall which is 0.6 metres thick and 10 and 14 metres deep was analyzed numerically using the Finite Element Method (FEM). Three stages of excavation were considered : first, complete construction of the Diaphragm Wall, second, 2-metres excavation without bracing, and finally excavated full depth with steel bracing. Clay was simulated by Elasto-Plastic Model (Mohr-Coulomb) and Modified Cam Clay Model. Sand is simulated by Elasto-Plastic Model (Mohr- Coulomb). Diaphragm Wall and steel bracing was simulated by using Elastic Model. The study focused on horizontal displacement, horizontal stress, ground settlement and bottom heave. The result showed that Diaphragm Walls of 10 metres and 14 metres in depth were similar in every behaviour pattern with difference in magnitude. Based on Modified Cam Clay Model, large magnitude was predicted when compared with Elasto-Plastic Model (Mohr-Coulomb). In stage 2, excavation of 2 metres in depth, Diaphragm Wall was translated and rotated about toe, while the tip was translated more than that observed by using inclinometer. Maximum horizontal displacement of Diaphragm Wall exhibited at 0.7 of depth of Diaphragm Wall at the finial stage. The horizontal stress predicted from the two studied Model was less than that proposed by PECK (1967). Modified Cam Clay Model showed more ground settlement and bottom heave than Elasto-Plastic Model (Mohr-Coulomb). In conclusion, behaviour of diaphragm wall by Finite Element Method using CRISP90 gives more magnitude than the actual field reading and has to be corrected by some parameters. Elasto-Plastic Model (Mohr-Coulomb) is more suitable for Chiang Mai Clay than Modified Cam Clay Model.