A method for trajectory optimization during the descent phase of helicopter autorotation is presented. The focus is on autonomous rotorcraft, although the approach could be extended to manned vehicles or used for pilot cueing. The approach is based on a Dubin's curve that has been modified to allow for acceleration and vertical translation along the path. A closed form solution to the accelerating Dubin"s path is presented and used to cast the problem of trajectory optimization as a parameter optimization problem. Path parameters (rotor speed, horizontal airspeed and bank angle) that bring the helicopter from the point of engine failure to a suitable state for safe flare and landing at a specified site can then be computed. Results of simulations involving autorotation to various final conditions are presented.