The experimental evidence which supports the existence of stable ductile crack extension in metallic specimens and structures is reviewed and compared with modern failure assessment procedures. The evidence includes data from the following: 1. Materials—aluminium alloys, mild steel, two nickel-chromium-molybdenum-vanadium forging steels, A533B and A508 pressure vessel steels, a quenched-and-tempered steel, austenitic steels, and HY-130. 2. Geometries—3-point bend specimens of various sizes, compact tension specimens plane and side-grooved, single-edge-notched tension specimens, double-edge-notched tension specimens, center-cracked tension panels, wide plate specimens, pressurized pipe geometries, and pressurized nozzle geometries with the crack located in the crotch corner, 3. Section sizes—thicknesses ranged from 2 to 700 mm in specimen geometries with crack sizes between 0.1 and 0.9 of the section.

It is concluded that for materials with a low to moderate capacity for strain hardening, an adequate engineering assessment can be obtained using a simple form of assessment such as the Central Electricity Generating Board's procedures, R6. However, for materials with a high capacity for strain hardening, total elastic plastic behavior can be predicted only by a procedure which allows for this strain hardening capacity explicitly. Nevertheless, the load capacity of such materials appears to be satisfactorily predicted using a flow stress criterion.