The application of peel tests for the measurement of adhesive fracture toughness of metal-polymer laminates is reviewed and the merits of a mandrel peel method are highlighted. The mandrel method enables a direct experimental determination of both adhesive fracture toughness (G(A)) and the plastic bending energy (G(P)) during peel, whilst other approaches require a complex calculation for G(P). In this method, the peel arm is bent around a circular roller in order to develop a peel crack and an alignment load attempts to ensure that the peel arm conforms to the roller. The conditions for peel arm conformance are thoroughly investigated and the theoretical basis for conformation are established. Experimental investigations involve the study of the roller size (radii in the range 5-20 nun are used), the peel arm thickness (varied from 0.635 to 2.0 mm) and the magnitude of the alignment load. In addition, the plane of fracture is studied since fractures can vary from cohesive to interfacial and this has a profound influence on the value of GA and on interpretation of results. A test protocol for conducting mandrel peel is developed such that the roller size for peel arm conformance can be established from preliminary fixed arm peel tests. The work is conducted on two epoxy/aluminium alloy laminates suitable for aerospace applications. Comparative results of adhesive fracture toughness from mandrel peel and multi-angle fixed arm peel are made with cohesive fracture toughness from a tapered double cantilever beam test. (c) 2006 Elsevier Ltd. All rights reserved.