<p>We investigate whether signatures of the underlying microstructure can be revealed through a post mortem statistical characterization of the fracture profile. To this end, we use the phase field model of fracture within a Finite Element framework to generate cracks that are resolved at the dominant microstructural length scale and run at least 100 times longer. The synthetic microstructures through which the crack propagates are carefully designed to provide some control over the eventual crack path. In each case, the fracture specimen is loaded in remote Mode-I. Cracks that are designed to be perfectly intergranular or propagate through a field of random toughness variations, lead to fracture profiles that are invariably flat at large scales. However, when profiled at the level of the dominant microstructural length scale, the same profiles appear to be self-affine and anti-persistent. On the other hand, in microstructures with randomly distributed defects that force the crack to follow a path largely dictated by the distribution of defects, the fracture profile is self-affine and persistent over a larger range of length scales. Thus, large scale persistence of the fracture profile in brittle fracture seems to be a sure indicator of the presence of random, ‘crack attracting’ defects like voids. In case of anti-persistent profiles, the microstructure is harder to discern. We show that even in these cases, the distribution of slopes of the crack segments, which in turn are connected to the local Mode-II perturbations encountered by the propagating crack tip, can provide some useful information about the nature of the underlying microstructure.</p>