Prediction accuracy of the two well-known anisotropic yield criteria, namely, Hill'48 and Yld2004-18p, on plastic deformation of AA5052-O sheet aluminium alloy experiencing compound stress was profoundly examined and contrasted in this work. To obtain the flow behaviour and anisotropy of the test material in varying loading directions, both uniaxial tensile and balanced-biaxial tests were conducted in the corresponding directions. Furthermore, a disc-compression test was carried out to reveal the biaxial r-value. Those gained mechanical properties could therefore be used to determine the material constants required by each yield criterion. The standard conical-punch hole expansion test was chosen here to impose complex stress on the sheet specimen. Forming parameters such as the punch load, stroke paths, hole expansion ratio, sheet thickness profile around the hole circumference and flange height were all monitored and investigated. It was shown that the results from the Yld2004-18p model with an exponent of eight apparently better agreed with the experimental data than those from the Hill'48 one. A conclusion could be drawn here that the yield criterion significantly affected the accuracy of the predicted deformation behaviour of the investigated aluminium alloy grade AA5052-O nevertheless in dependence on the degree of material anisotropy.