Creep rupture strength of creep strength enhanced ferritic steels is often overestimated, and its evaluated value has been reduced repeatedly. In this paper, the cause of the overestimation is discussed, and the creep rupture strength of T91 steel is assessed with its updated creep rupture data. Effects of residual Ni concentration on the creep rupture strength and necessity of F factor in T91 steel are also discussed. Decrease in activation energy Q for rupture life in long-term creep is the cause of the overestimation, since conventional time–temperature parameter (TTP) methods cannot deal with the change in Q. Due to the decrease in Q, long-term creep rupture strength evaluated decreases as longer-term data points are added or shorter-term data points are discarded in the conventional TTP analysis. The long-term region with small values of activation energy and stress exponent is named region L2 in this paper. Region L2 appears in all the heats of T91 steel and plate products of Gr.91 steel. Since service conditions of the T91 steel are usually in region L2, the creep rupture strength under the service conditions should be evaluated from the rupture data in region L2 only. The 5 × 105 hrs rupture strength at 550 °C decreases from 129 MPa (evaluated from the whole data of T91 steel) to 79 MPa (evaluated from the data in region L2 only) with increasing cut-off time for data selection. The 105 hrs rupture strength at 600 °C also decreases from 87 MPa (whole data) to 70 MPa (region L2 only) despite sufficient number of long-term data points at 600 °C. Careful consideration on the data selection is necessary in evaluation of creep rupture strength of the T91 steel. A multiregion rupture data analysis (MRA) is helpful to select data points belonging to region L2.