With regard to the quantitative modelling of delayed hydride cracking (DHC), it has been shown in Part 1 [9] that when a lenticular shaped hydrided region, i.e. one whose length is large compared with its thickness, forms at a planar surface or at the surface of a very blunt notch, the compressive stress sigma(H) induced within the region is markedly influenced by the unconstrained transverse precipitation strains as well as the unconstrained normal strain. The sigma(H) values are approximately the same irrespective of whether we assume that (a) the overall unconstrained expansion strain associated with hydride precipitation is confined entirely to the norm a I direction or (b) the strain is partitioned approximately equally between the three orthogonal directions. Thus, assuming the strain to be entirely in the normal direction allows for both precipitation strain scenarios. The paper extends the considerations in Part I to the case where there is a hydrided region immediately ahead of a sharp crack, i.e. the other extreme to that considered in the earlier work, a model situation that simulates the behaviour of a growing DHC crack. In this case the normal stress within the region immediately ahead of the crack tip is not compressive but tensile, and is influenced by the unconstrained transverse precipitation strains, though not to the same extent as is the hydride induced stress associated with a hydrided reg ion that ema nates from a planar surface. Assuming the strain to be entirely in the normal direction overestimates the local stresses, and therefore unlike the planar surface situation, the assumption does not allow for both precipitation strain scenarios. It is therefore important to input the correct unconstrained precipitation strain tensor, if we require a reasonably accurate quantitative picture of DHC crack growth.