The goal of the study was to analyze fracture properties of adhesive bond using a three-point end-notched flexure test and the compliance-based beam method. Critical strain energy release rates (GIIc) and cohesive laws were obtained for adhesive bonds made of European beech (Fagus sylvatica L.) and adhesives such as EPI, MUF, PRF and PUR. The experiments were assisted with FE analyses employing three different material models of wood: elastic (Elas), symmetric elasto-plastic (EP) and elasto-plastic with different compressive and tensile yield stresses parallel to fiber (EP+). The highest mean GIIc was achieved for PUR (5.40 Nmm−1) and then decreased as follows: 2.33, 1.80, 1.59 Nmm−1 for MUF, EPI, and PRF, respectively. The failure of bondline was brittle and occurred at bondline for EPI, MUF and PRF, and ductile and commonly occurring in wood for PUR adhesive. The FE simulations employing cohesive models agreed well with the experimental findings for all adhesives. FE model with Elas material was found accurate enough for EPI, MUF and PRF adhesives. For PUR adhesive, the model EP+ was found to be the most accurate in prediction of maximal force. The impact of friction between lamellas may be up to 4.2% when varying friction coefficient from 0 to 1. The impact of the grain angle distortion (α) with respect to longitudinal specimen axis showed its high influence on resulting stiffness and maximal force. It was found that three-point end-notched test is suitable for EPI, MUF, and PRF, while it is less appropriate for a bond with PUR adhesive due to notable plastic behavior.