The effects of small artificial defects on the fatigue limit of an annealed carbon steel have been investigated in the temperature range of 20°C to 450°C. The experimental results have been compared with those predicted by Murakami's √(area) model. Small drilling holes were used as artificial defects. The diameters of the holes were 0.1, 0.2 and 0.3 mm, respectively and their depths were almost the same as the diameter. The fatigue limits of the drilled specimens as well as the smooth specimens showed a sharp peak caused by dynamic strain aging near 350°C. Vickers hardness, HVs, necessary for the application of the √(area) model was measured in the same temperature range as the fatigue tests. The Vickers hardness, HVs, did not show such a sharp peak as the fatigue limit near 350°C. The fatigue limits obtained from the fatigue tests agreed with those predicted by the model within a scatter band of ±10% at 20°C and 250°C. The experimental values, however, were larger than the predictions at 350°C and 450°C. At 350°C, the deviation was large and its degree became about 35%. The results were attributed to a much greater effect of dynamic strain aging for the fatigue limit than for the hardness. When the hardness, HVc, measured after fatigue loading was applied to the model, the deviation was decreased to 18%. Thereafter modified hardness, HVf, obtained by multiplying fatigue limit ratio, Rf, to the hardness, HVs, was applied to the model, where Rf is the ratio of high temperature fatigue limit to room temperature fatigue limit. Most of the results were in the scatter band of ±10% from the predicted value.