When using the plastic limit load concept for structure failure analysis, the stress analyst must confirm that the material of the structure possesses adequate ductility to deform in the plastic domain without instability, even in the presence of known or assumed manufacturing defects. The work reported in this paper has determined that, for use with a lower-bound plastic limit load method, this minimum adequate ductility is 45 J, expressed in terms of Charpy-V-notch energy ( C v-energy ) as measured by standard material tests. The lower-bound method uses plane stress formulae and the yield strength in simple tension σ y instead of the flow stress σ∗ (i.e. no accounting for strain hardening) to yield results on the safe side. This determination is based on an evaluation of more than 400 burst and fracture mechanics tests on components (shell type structures) and specimens (various tensile and bend specimens) made of ferritic and austenitic steels of nuclear grades as well as conventional structural steels, and including some intentionally degraded melts. Another proposed criterion, namely that the C v-energy to yield strength ratio should be equal to or greater than 0·12 J MPa −1 for the application of the plastic limit load concept, was found to result in an unnecessarily stringent limit for materials above 375 MPa yield strength. In addition, the reduction in area measured in tensile tests has been investigated as an alternative index of toughness. This property, however, was found to be unsuitable for defining a ductility limit for use with the plastic limit load concept. Therefore, the aforementioned easy-to-use engineering rule, which requires only geometrical data and two material properties, C v-energy and the yield strength in simple tension, is recommended for failure analysis of cracked structures.