Available acceptance and rejection criteria for discontinuities in steel components of a welded assembly focus only on a minimum limit for the steel fracture toughness and a maximum limit for the size of the discontinuity. However, other critical parameters such as the uncertainty in the fracture toughness value and the working stress type in the steel component are not taken into account. The uncertainty in the fracture toughness increases as the steel component becomes thicker, which in combination with an increase in the demand for thick steel components in construction is cause for concern. Proposed in this paper is a probabilistic approach for the assessment of discontinuities that accounts for the variability of the fracture toughness throughout the material as well as the working stress type. A database of CharpyV-notch (CVN) impact test results was developed for steel plates of different grades, thicknesses and at different temperatures. Each dataset was fitted with a statistical distribution, which was then converted to the corresponding fracture toughness. Through logistic regression and linear fracture mechanics an expression was developed to calculate the probability of a discontinuity’s stress intensity factor exceeding the fracture toughness of the material. By determining an acceptance limit of this probability value based on the working stress type of the component and assembly (importance) the proposed approach can be implemented to accept or refuse a discontinuity in a steel component.