Self-healing of creep-induced damage in newly designed 12Cr self-healing ferritic steels is studied. The damage healing is achieved by the segregation of supersaturated solute atoms at the free surface of the creep-induced cavities, and in this research, the healing phase is the W-riched Laves phase. Two kinds of self-healing steel with different precipitation driving forces of Laves phase are investigated, namely fast self-healing steel (FSHS) and slow self-healing steel (SSHS), in which the FSHS is designed to show the healing effect after 280 hours, while SSHS is designed to show healing effect after 27000 hours. Although both systems are self-healing systems, only FSHS is expected to show the self-healing phenomenon under the testing condition of this study, while SSHS not. Creep tests were performed on bone-shaped samples with constant stresses ranging from 100 to 260 MPa at 550 °C. The creep behavior is compared with traditional 9-12Cr ferritic steels, the relationship between the minimum creep rate and stress obeys Norton's power law, and the relationship between the lifetime and stress also obeys a power law. The total strain and stress exponents in power law of both SSHS and FSHS are relatively high compared to traditional 12Cr ferritic steel with similar grain size. The microstructure of the creep-failed samples, including the fracture surfaces, the uniformly-strained region (i.e., the stress-affected region), and the stress-free region are investigated with scanning electron microscopy (SEM). By comparing the feature of the stress-affected region and stress-free region exposed to the same thermal history, the effect of high-temperature exposure could be isolated from this. The effect of stress on the Laves phase precipitation behavior and the damage evolution could be understood more clearly. For both the SSHS and FSHS, a ductile and transgranular fracture mode is confirmed by analyzing the fracture surfaces. In the stress-affected region and stress-free region of the creep-failed samples, M₂₃C₆ precipitates, Si/C-rich impurities, and pores were found in all observed samples. The Laves phase was only found in the samples with a longer lifetime (> 300 h). A TEM observation and EDX scanning was conducted on the FSHS sample with a lifetime of 2487.2 hours. Laves phase precipitates were found on the grain boundaries and in the matrix, while no grain boundary cavities were found. Statistics based on pores observed on samples tested under different stress showed that the pores were present prior to creep tests and are generally not due to creep. A combined analysis of creep behavior and microstructure showed that dislocation climbing is likely the dominant creep mechanism. The absence of typical creep grain-boundary cavities was allocated to the relatively ductile matrix of 12Cr self-healing steel or the unfavorable dominating creep mechanism for creep cavity forming. With the prerequisite of self-healing not found in all tested samples, self-healing was considered not found under current testing conditions. The statistics based on the Laves phase showed that the precipitation depends on creep time, and the stress is indeed lowering the nucleation barrier, thus promoting the Laves phase to precipitate in the stress-affected areas.