A scaling approach, based on the deformation J-integral at maximum load obtained from small specimens, is proposed for predicting the crack instability behaviour of burst tests on irradiated Zr-2.5Nb pressure tubes. An assessment of this approach is carried out by comparison with other toughness criteria such as the modified J-integral and the plastic work dissipation rate approach. The largest discrepancy between the different parameters occurs for materials of intermediate toughness which exhibit the most stable crack growth and tunnelling up to maximum load. A study of one material of intermediate toughness suggests crack-front tunnelling has a significant influence on the results obtained from the 17-mm-wide specimens. It is shown that for a tube of intermediate toughness the different approaches can significantly underpredict the extent of stable crack growth before instability in a burst test even after correcting for tunnelling. The usefulness of a scaling approach in reducing the discrepancy between the small- and large-scale specimen results for this material is demonstrated.