An exploratory experimental study has been made of those bond strength test methods that are amenable to a fracture mechanics interpretation: peel, rod pull-out, and simple shear. Equations for calculating fracture energies from these test pieces are given. For strong bonds, the calculated fracture energies are not independent of the test geometry. This is attributed to different morphologies of the failure surfaces: the sharper the effective crack tip, the lower the fracture energy. Fracture surfaces observed for peel at low angles (near-bond failure) and for simple shear (failure at the bond or in the rubber leaving smooth fracture surfaces along a straight trajectory) are taken to correspond to sharp crack tips, in contrast to the rough fracture surfaces formed in the rubber for the other test geometries. Application of fracture mechanics to the simple shear test piece is complicated by the need ideally to use the retraction energy in the calculations, and by the observation that failure did not initiate from artificially introduced cuts placed where intuition suggests initiation should occur.