This work investigated accuracy of various mathematical expressions used to calculate the critical strain energy release rate ( G c) for delamination in fibre composites. Three mathematical expressions were considered here, based on (i) a simple beam theory, (ii) a transverse shear deformation theory, or (iii) a corrected beam theory with consideration of transverse shear deformation and crack tip singularity. Variable selected to examine accuracy of these expressions was specimen thickness. Since G c is a material property, change of specimen thickness should not affect its value. The study used 2-dimensional finite element models with a blunt starting defect, which have length and geometry simulating the test coupons used for the delamination tests. For delamination in the shear mode (Mode II), we assumed that contact surfaces along the starting defect were free from friction, in order to be consistent with the beam theory expressions used for the calculation of G c. As the finite element analysis used is static in nature, only the strain energy release rate for crack initiation was examined. The study firstly assigned a constant load of 1 N for the 10 mm-thick models, and then calculated the corresponding loads for models of other thickness based on constant strain energy release rates, G I and G II for Mode I (tension mode) and Mode II respectively, using the three beam theory expressions. For each model under the given load, stresses in the vicinity of the starting defect were then examined to determine whether the specimen thickness affects the stress values. Stresses used were the maximum principle stress and the von Mises stress along the contour of the starting defect, and the normal stress and shear stress along the boundary of the interlaminar resin-rich region, which were treated as the stress criteria for fracture initiation. The study concludes that the corrected beam theory provides G c expressions that are least sensitive to the specimen thickness in both deformation modes.