Acoustic streaming is a key factor in ultrasonic casting of aluminum alloys because it determines heat and mass transfer, and thus affects the temperature distribution, solidification phenomena, solute transport and crystal growth. However, measurement and prediction of acoustic streaming in molten aluminum present a big challenge due to high temperatures and acoustic streaming complexity. In this work, a numerical model has been developed to simulate acoustic streaming in molten aluminum. The model allows predicting generation of cavitation bubbles, sound wave propagation and attenuation due to interaction with the bubbles, and fluid flow. The model validation has been performed by comparing the predicted and measured acoustic streaming in water. Good agreement between these results suggests that the model can be used to predict acoustic streaming in actual casting processes. However, a number of issues still remain to be solved particularly an accurate simulation of bubble oscillations.