In this paper, a developed energy based multiscale damage criterion is investigated for biaxial loading case. This criterion incorporates crystal plasticity at the microscale and produces a damage tensor which represents the local damage state deriving from least square method. The damage tensor driven by a modification of strain energy density on each potential slip system is averaged from local to grain level to obtain a damage vector for each grain. Subsequently, in order to predict the failure of a meso representative volume element (RVE), the Kreisselmeier-Steinhauser (KS) function, which can produce an envelope function for multiobjective optimization was adopted to calculate the damage index for meso RVE. Weighted averaging method was also used to provide most potential cracking directions for meso RVE at the same time. In order to verify that the developed method is capable to give an acceptable prediction of fatigue damage initiation and growth under multiaxial loading condition, a cruciform specimen was used for biaxial loading. A biaxial torsion MTS machine was used to carry out fatigue tests on the cruciform specimen. Numerical fatigue analysis was also performed based on the multiscale fatigue damage criterion. Comparing the simulation results with the experimental data showed that the multiscale fatigue damage model can provide acceptable prediction on failure of meso RVE and the crack direction.