This paper addresses the narrow applicable fracture toughness range of the master curve method (ASTM-E1921) by expanding the scaling process to include the effect of constraint loss in miniaturized specimens. To facilitate the acquisition and scalability of reliable fracture toughness data from one specimen size to another and to apply the Master-Curve method with confidence when using small specimens, the difference in the structure of the crack tip stress field between large and small specimens has to be well understood and modeled. In this study, the crack tip stresses of two specimen sizes (1T and 0.18T compact tension specimens) were calculated with 3D finite element simulations run over a range of loadings (or equivalently for a range of constraint levels) and at different temperatures in the lower part of the ductile to brittle transition of a structural steel. The toughness scaling, based on a local approach to brittle fracture, is compared with the pure statistical correction of the crack front length recommended in the ASTM standard E1921. We show that the effect of constraint loss already becomes significant at low K_Jc values. Consequently, we propose a simple “modified correction factor”, k_M, based on the ratio of crack front length and critical area/volume, accounting for constraint loss. Applying this correction factor, we show that scaling between specimen sizes and geometries yields very good correlation between simulated and scaled/calculated stress intensities.