The use of a fracture mechanics approach for design of engineering structures is relatively rare compared to stress or stiffness based approach. Using a fracture mechanics approach, the influence of defects and their propagation can be better described than using a stress or stiffness approach. Also, stress or stiffness approaches are generally conservative compared to the fracture mechanics approach. Despite these advantages, a fracture mechanics approach is relatively rarely used in engineering design due to the difficulty to obtain accurate energy release rates (G-value) and to the derivation of the compliance/crack length equation. In this study, the energy release rate value was derived based on a standard 3 points end notched flexure samples (3-ENF) for a crack propagation in mode II. To investigate whether the determined energy release rate (ERR) value is influenced by the sample size, a new sample geometry was designed with a crack positioned in the middle of the specimen (but without open cracked ends) and compared with the 3-ENF specimen. For this new geometry, the compliance/crack length equation was determined according to the Griffith approach. The specimen compliance was derived according to the Euler-Bernoulli equation. The specimen was tested under quasi-static and fatigue flexure loads. It was shown that the fracture mechanics approach could not simply be used on this new sample geometry. Two possible explanations are discussed, one being the difficulty of obtaining ERR-values which are not influenced by the sample geometry, and the limitation of using a Euler-Bernoulli approach to derive the compliance of specimen. This study presents and discusses a few of the major difficulties which influence the application of fracture mechanics approach to design of engineering structure, such as influence of the sample size on ERR value and the validity of the Griffith equation for complex structures and anisotropic materials.