The use of adhesive joints in industrial applications has been increasing. To increase the confidence in the design of adhesive structures, it is important to be able to accurately predict their strength and fracture properties (critical strain energy release rate in tension, \(G_{\mathrm{IC}}\), and shear, \(G_{\mathrm{IIC}})\). As in most cases loads induce mixed-mode (tension plus shear), it is of great importance the perception of fracture under these conditions, in which relates to the strain energy release rate in tension, \(G_{\mathrm{I}}\), and shear, \(G_{\mathrm{II}}\). This analysis allows choosing the best failure criterion to use in cohesive zone models, to predict the behaviour of bonded joints. This work presents an experimental and numerical study using the Single-Leg Bending (SLB) test on specimens bonded with three adhesives, to study and compare their mixed-mode fracture properties. For this purpose, data reduction methods were applied to estimate \(G_{\mathrm{I}}\) and \(G_{\mathrm{II}}\) that require the measurement of crack length (a) and methods using an equivalent crack length (\(a_{\mathrm{eq}})\). The analysis and comparison of \(G_{\mathrm{I}}\) and \(G_{\mathrm{II}}\) obtained during the experimental phase were addressed. Framing the obtained values in several fracture envelopes enable to select which failure criterion is more appropriate for each adhesive. Overall, a very good agreement was obtained between methods for the determination of \(G_{\mathrm{I}}\) and \(G_{\mathrm{II}}\). Actually, apart from one of the methods that produced higher errors, the highest deviation to the \(a_{\mathrm{eq}}\)-based method (considered as the most robust) was 9.3%, obtained for \(G_{\mathrm{I}}\). In the numerical simulations it was possible to reproduce the observed behaviour of the experimental tests, with a positive validation of the chosen propagation criteria obtained from the experimental results.