This paper describes the accuracy of FEM analysis for the estimation of local strain at the notch root in creep-fatigue conditions, comparing the local strain calculated by FEM with the measured strain using high temperature strain gages. The specimens are commercial pure aluminum notched plates with both side grooves whose elastic stress concentration factors are 0.5 and 2.0. The test was made at 473 K in air. The following results are obtained. In no-hold time fatigue tents, FEM analysis accurately estimates the local strains at the notch root. In no-hold time fatigue tents, FEM analysis accurately estimates the local strains at the notch root. In the hold time test, on the other hand, FEM analysis estimates a smaller strain range compared with the actual strain range occurring at the notch root. The reason for the disagreement between FEM analysis and the actual strain results from the change in material response due to creep loading. In the creep-fatigue condition, so-called plastic-creep interaction, the FEM analysis does not take account of, occurs. The local strain calculated from conventional methods, Neuber's rule, Stowell-Hardrath-Ohman, Koe's methods are also compared with the strain measurement for no-hold time tents. All the methods overestimate the strain concentration at the notch root.