Deep excavations constructed rapidly in soft clay with poor drainage performance require resistant heave stability analysis under undrained conditions. Currently, total stress analysis with the total stress strength indices of conventional triaxial consolidated-undrained tests (Method C) is commonly used in domestic engineering community. In contrast, the academic and foreign engineering community recommend the effective stress analysis method based on effective stress strength indices (Method A) and the undrained strength analysis method based on undrained shear strength (Method B). Combining with real excavation cases, the applicability and rationality of the three methods for analyzing excavations are studied by the finite element limit analysis. The results show that: As the total stress path in excavation differs from that of the conventional triaxial compression test, Method C underestimates the true shear strength of unloaded soil, while CU tests overestimates the in-situ undrained shear strength. Therefore, the calculated safety factor by Method C is not reliable. Method A overestimates the undrained shear strength of shear-shrinkage clay and the safety factors. Method B is the most accurate theoretically.