This report presents the fracture energy and the complete stress – elongation relation for a structural adhesive loaded in modus I. The experiments are performed on the Double Cantilever Beam (DCB) specimen and the method to analyse the experiments is based on the J-integral approach which means that the energy release rate, i.e. J, is measured continuously during an experiment. Since J is given by the area under the stress – elongation relation for the adhesive layer, both the fracture energy and the stress – elongation relation can be measured in the experiments. The geometry of the specimens is varied in order to examine if the evaluated stress – elongation relation is a unique constitutive relation for the adhesive layer. No dependence on the specimen geometry has been detected provided that the adherends only deform elastically. If the adherends are allowed to deform plastically the fracture energy increases and the stress – elongation relation from the elastically deforming tests cannot be used to simulate the structural behaviour. This is interpreted as an effect of a substantially shorter damage zone with the plastically deforming adherends. An evaluation of effects of the loading rate is also performed. This shows that the fracture energy increases with the loading rate. A number of standardized methods are available to evaluate the fracture properties of adhesives from experiments on the DCB-specimen. Furthermore, alternative methods have recently been suggested by Tamuzs et al. (2003). In an effort to examine the accuracy of the alternative methods, FE-simulations are performed and evaluated according to the alternative methods. The simulations are based on the stress – elongation relation evaluated from the present experiments. Large differences between the correct and the evaluated fracture energies are found for most of the methods. Two of the methods show good accuracy and appear particularly promising. Stress whitening is used to evaluate the length of the damage zone. This length compares reasonably well with the length evaluated from FE-simulations of the experiments.