Wedge tests are usually analysed assuming that the free, unbonded members may be treated as encastre cantilever beams. However, if the adhesive layer is sufficiently flexible (e.g., due to low elastic modulus), then significant strain in the bonded region may occur and lead to modification of the behaviour outside this region. Using in conjunction a sensitive strain gauge method on asymmetric wedge tests and a mathematical analysis developed from the work of Winkler, we conclude that the standard, simple beam theory approach significantly overestimates crack length for a supple adhesive layer. The present contribution mainly considers strain effects in the intact, bonded zone, rather than fracture per se. However, it is concluded that, if in fracture tests, the incorrect values of crack length obtained from the encastre beam assumption are employed to calculate fracture energy using the simpler model, there will be some self-compensation and little error in estimates of the latter will result (at least in the cases presently studied). (C) Koninklijke Brill NV, Leiden, 2011