In this talk I review two current problems where statistical mechanics and fracture meet (see Adv. Phys. 55, 349 (2006).  The first is the so-called size effect, which measures the strength of a sample as a function of its size.  This can be addressed both in the traditional FSS sense - becoming an extremal statistics problem - and in the presence of one defect, a notch (Phys. Rev. Lett. 100,  055502 (2008)).  I present some recent results on the application of scaling theory and extensive simulations of simple fracture models to illustrate the fundamental role of disorder here.  The second problem is the creep of a fracture line. I discuss experiments done at HUT using the so-called peel-in-nip geometry to split paper in two (Phys. Rev. Lett. 99, 145504 (2007)) in the creep setup, or under a constant external force.  The scenario can be used to test fundamental results in the theory of elastic manifolds in random media, in particular the depinning of lines with long-range elasticity.  Listening to the dynamics via acoustic emission measurements allows to look at the temporal dynamics with high accuracy.