Abstract The paper aims at contributing to a better appraisal of the wedge-opened double cantilever beam method for testing the debonding toughness of adhesively bonded plates. The test specimens consisted of two plates of either hot-dip galvanized low carbon steel or alloy Al-6082-T6 bonded with an epoxy-based adhesive. The classical method of crack length measurement by visual observation of the crack tip along the side of the specimen is compared with a method making use of displacement sensors for continuously monitoring the deflection of the plates. The variation of toughness with debonding rate was derived from the evolution of the debond length either with respect to a static wedge or with respect to a wedge advancing at a constant rate. The paper enlightens the uncertainty arising from the anticlastic effect and compares the experimental reproducibilities provided by the different test procedures.