本研究擬建立一套極值統計方法，分析軸承鋼料中非金屬介在物最大尺寸及類型，並探討鋼料中不同取樣部位、試片數目及介在物類型對極值統計預測曲線之影響。結果顯示鑄模受冷卻循環作用與鋼液擾動產生漩渦作用而易附著於接近凝固殼邊緣處之影響，造成 D/4 部位之內部介在物的最大尺寸會比 D/2 部位較小且分佈較均勻。此外，介在物的顯微型態亦會直接影響 SEV預測曲線，若複合較多型非金屬介在物（如 MgO.CaO.Al2O3型）斜率值較低，並可預測出鋼中內部介在物尺寸會較大。若能在煉鋼技術上能避免含有三種以上的氧化物複合再一起時，其介在物之尺寸即可降至約 16 μm以下，方能進一步提升軸承鋼料的品質及改善其滾動疲勞壽命。

The fatigue life of the bearing steel was estimated by the statistics of extreme values (SEV) method in this study. Various sections of center (D/2) and side (D/4) sections of the bearing rod were examined by the prediction of the SEV method. The superior cleanliness of this steel was obtained in the side D/4 section from the SEV prediction. The compound nonmetallic inclusion, i.e., MgO˙CaO˙Al2O3, was examined to be the largest inclusion type in both the D/2 and D/4 sections of this bearing steel. The inclusion types in the bearing steel are discussed. In addition, the reason resulted in the various fatigue life in the various sections are demonstrated.