Path following control is the basis for autonomous navigation of underactuated unmanned surface vehicles (USV), in which the line-of-sight (LOS) guidance algorithm is widely employed due to its computational efficiency and reliability. However, the LOS guidance law only considers the vehicle kinematics, as well as the saturation constraints of the vehicle actuators, and the vehicle thereby experiences steering lag while tracking a large curvature path, resulting in large tracking errors. Therefore, this paper proposed a LOS guidance algorithm augmented with feedforward compensation to address this problem. The lookahead distance of LOS guidance is compensated dependent on the steering performance and the curvature of the desired path. When the steering performance of the ship is worse or the curvature of the path is greater, the forward looking distance will be decreased and the desired course angle will be increased. In this way, the vehicle can reduce the steering lag caused by the saturation constraint. The accuracy of curve path tracking is evaluated for traditional LOS guidance, integral LOS guidance, and the proposed feedforward LOS guidance algorithm by using numerical analyses. The results demonstrate that the proposed feedforward LOS guidance exhibits the lowest lateral error during curve path tracking. It indicates that the proposed feedforward guidance algorithm can effectively reduce the steering lag of the vehicle to improve the tracking accuracy.