EN 1993-1-10 contains tables for selection of steel quality class with respect to avoidance of brittle fracture. The conservatism of the method has not previously been examined statistically. The original verification of the method has been established basically on a deterministic assessment of 19 double edge cracked specimen tests. This verification has not accounted for e.g. constraint effects and the treatment of nominal data versus measured data is not fully satisfactory. Therefore, here the confidence of the procedure is examined statistically using more up to date information regarding the Charpy-V vs fracture toughness correlation and a quantitative assessment of the constraint effect. The study has two main purposes. First, to present a method by which consistent statistically defined criteria for a standard like EN 1993-1-10 can be developed and second, to obtain an approximate estimate of the conservatism contained in the present methodology. The statistical treatment is applied to the same 19 specimens originally used for the verification of the method.