In the paper notch toughness assessment of full scale testing samples (FS) form the upper bound toughness of sub-sized (SS) samples of structural carbon-manganese steels. The relations proposed by Schindler (2000) are in good agreement with experimental data. Empirical proportionality constant q* = 0,54 between notch toughness of full scale and sub-sized samples of studied structural steels agrees well with theoretically estimated constant q* = 0,50–0,54. More precise knowledge of the size effect of testing samples on temperature dependence of notch toughness requires an analysis of scatter in experimental data.