The so-called EURO data set is the largest set ever assembled, consisting of fracture toughness results obtained in the ductile-to-brittle transition region. It was the outcome of a large project, sponsored by the European Union, which involved ten European laboratories in the second half of the 1990s. Several post-project investigations have identified one of the blocks from which specimens were extracted (block SX9) as macroscopically inhomogeneous and significantly tougher than the remaining blocks. In this paper, the variability of block SX9 has been investigated using the conventional master curve (MC) methodology and some recent MC extensions, namely, the SINTAP (structural integrity assessment procedure) lower tail, the single point estimation, the bi-modal MC, and the multi-modal MC. The basic MC method is intended for macroscopically homogeneous ferritic steels only, and the alternative approaches have been developed for the investigation of inhomogeneous materials. Therefore, these methods can be used to study the behavior of block SX9 within the EURO data set. It has been found that the bi-modal and multi-modal MC approaches are quite effective in detecting the “anomaly” represented by block SX9 but only when analyses are performed on data sets of comparable size.