A B S T R A C T Different laboratory experiments are usually conducted to characterize the fracture behaviour and integrity properties of newly developed structural materials. However, the reported fracture tests data for an improved high strength cement mortar (HSCM) under combined tension-shear loading are not in agreement with theoretical predictions obtained from well-known fracture criteria. It is shown in this paper that the significant difference existing between the experimental and theoretical results is due to ignoring the effect of T-stress on the processes of crack growth in the HSCM test specimens. A modified fracture model is then used to show that the theoretical predictions can be corroborated by the experimental results when the effect of T-stress is taken into account.