Historically the ASME reference curves have been treated as representing absolute deterministic lower bound curves of fracture toughness. In reality, this is not the case. They represent only deterministic lower bound curves to a specific set of data, which represent a certain probability range. A recently developed statistical lower bound estimation method called the ‘Master curve’, has been proposed as a candidate for a new lower bound reference curve concept. From a regulatory point of view, the master curve is somewhat problematic in that it does not claim to be an absolute deterministic lower bound, but corresponds to a specific theoretical failure probability that can be chosen freely based on application. In order to be able to substitute the old ASME reference curves with lower bound curves based on the master curve concept, the inherent statistical nature (and confidence level) of the ASME reference curves must be revealed. In order to estimate the true inherent level of safety, represented by the reference curves, the original database was re-evaluated with statistical methods and compared to an analysis based on the master curve concept. The analysis reveals that the 5% lower bound master curve has the same inherent degree of safety as originally intended for the KIC-reference curve. Similarly, the 1% lower bound master curve corresponds to the KIR-reference curve.