Standard procedures for a fracture toughness testing require very severe restrictions for the specimen geometry to eliminate a size effect on the measured properties. Therefore, the used standard fracture toughness data results in the integrity assessment being irrationally conservative. However, a realistic fracture in general structures, such as in nuclear power plants, may develop under the low constraint condition of a large scale yielding with a shallow surface crack. In this paper, cleavage fracture toughness tests have been made on side-grooved PCVN (precracked charpy V-notch) type specimens (10 by 10 by 55 mm) with various crack depths. The constraint effects on the crack depth ratios were evaluated quantitatively by the developed scaling method using the 3-D finite element method. After the fracture toughness correction from scaling model, the statistical size effects were also corrected according to the standard ASTM E 1921 procedure. The results were evaluated through a comparison with the $T_0$ of the standard CT specimen. The corrected $T_0$ for all of the PCVN specimens showed a good agreement to within $5.4^{circ}C$ regardless of the crack depth, while the averaged PCVN $T_0$ was $13.4^{circ}C$ higher than the real CT test results.