Prediction of the maximum size of inclusion in a large weight of steel using data from a small volume steel sample is an important topic for steelmakers. Therefore, the probable maximum sizes (PMS) of inclusions in three steel grades were evaluated by the statistics of extreme values (SEV) and the particle size distributions (PSD) methods based on three-dimensional (3D) investigations of inclusions using the electrolytic extraction (EE) method. The effect of number of measurements and size of unit area on the PMS of in-clusions were investigated. The results showed that at least 80 measurements of NMIs should be done in the SEV method, while in the PSD method the number of measurements has little influence when the number of inclusions in the observed areas was large enough. The effect of unit area size on the PMS of inclusions in the SEV method can be ignored for small-sized inclusions (less than 10 mm). The PMS of inclusions determined from the SEV method agreed with that from the PSD method by considering the 95% confidence interval. The SEV method was preferred when predicting the PMS of inclusions because of its simplicity by reducing the cost and labour involved compared to the PSD method.