The single-step optimal control of a voltage-driven variable-reluctance stepping motor, with an inertia load, is considered. A solution for the optimal control policy is achieved by a conjugate gradient method in function space, optimising the response of a well established nonlinear motor model. The open-loop bang-bang policy which results is refined using a switching-times algorithm which is more economical in computer requirements. The technique is implemented as a microprocessor controller and is shown, by experimental results on a typical motor, to be effective over a wide range of load inertia.