This paper deals with a one-versus-one air combat algorithm that considers the direction of the lift vector of an unmanned combat aerial vehicle(UCAV). The air combat situation discussed in this study assumes that a gun-fight takes place within visual range(WVR), thus the air combat algorithm will perform the maneuver to chase the enemy's tail. The air combat algorithm previously proposed by the authors uses the score function matrix to predict the enemy's maneuver during the prediction time, and then selects the optimal maneuver. However, the previously proposed algorithm has a disadvantage in that the performance difference occurs due to a change in the prediction time. In order to overcome this disadvantage, we improve the existing algorithm by considering the direction of the lift vector. We perform combat simulations for several scenarios using a pseudo 6DOF model of the F-16 fighter to verify the performance of the air combat algorithm proposed in this paper.