Long-term rupture data of 11Cr-2W-0.4Mo-1Cu-Nb-V steel were analyzed using an exponential equation for stress regarding time to rupture as a thermal activation process. The fitness was compared with the usually employed method assuming power-law creep. In the exponential method, rupture data are classified into several groups according to the thermal activation process; the activation energy, Q; the activation volume, V; then, the Larson-Miller constant, C, values are calculated, and a regression equation is obtained for each data group. The fitness level of the equation was satisfactorily high for each group. The values of Q, V, and C were unusually small for a data group where an unexpected drop in rupture strength was observed. The critical issue is how to comprehend signs of degradation within the short term. We can observe several signs at a creep time of approximately one-tenth of the times of the degradation events. The small values of Q and V indicate that completely softened regions form and creep locally, which is consistent with previous observations. From both metallurgical considerations and the variations of Q and V, it is suggested that the rate of the unexpected drop in strength is mitigated after further long-term creep.