Brittle fracture in ferritic steels is a field in which a lot of research has been performed over the years. Most of this research has been done on uniaxially loaded specimens, such as single edge notched bend specimens. From these experiments, the maximum principal stress criterion is used in the Weibull distribution method as proposed by Beremin, and is observed to accurately predict the fracture toughness distribution. However, when multiaxial specimens are considered, the maximum principal stress criterion no longer accurately predicts the fracture toughness distribution. In this thesis, other failure criteria than the standard maximum principal stress are considered as a solution to this problem. Using the data provided in the 2006 paper "An Experimental Investigation of the Effect of Biaxial Loading on the Master Curve Transition Temperature in RPV Steels" by R. Link, A Joyce and C. Roe, the properties of Shoreham pressure vessel steel are obtained. Furthermore, the fracture toughness from the tested cruciform specimens allows for the reconstruction of the stress state around the crack tip during fracture. This is done by creating various cruciform specimens in the Abaqus finite element analysis program, which are loaded in accordance with the paper. The resulting stress state is used in the calibration of the Weibull parameters in the Weibull distribution. It was found that uniaxially loaded specimens show good agreement with the predicted failure probabilities. Additionally, it was confirmed that the biaxially loaded specimens do not show good agreement when the maximum principal stress is the failure criterion. When the failure criterion is altered so that only microcracks that do not experience large triaxiality contribute to fracture, good agreement is obtained for both the uniaxially and biaxially loaded specimens. Hence, it is found that triaxiality is very important for cleavage fracture, with high levels of triaxiality preventing microcracks from propagating. This leads to a proposed failure criterion where the maximum principal stress criterion is applied, and only elements that do not experience high levels of triaxiality contribute. It is suggested to further test these conclusions under a variety of loading conditions, for which an alternate specimen is proposed.