Ancillary services are necessary for maintaining the security and reliability of power systems and constitute an important part of trade in competitive electricity markets. Spinning Reserve (SR) is one of the most important ancillary services for saving power system stability and integrity in response to contingencies and disturbances that continuously occur in the power systems. Hence, an accurate day-ahead forecast of SR requirement helps the Independent System Operator (ISO) to conduct a reliable and economic operation of the power system. However. SR signal has complex, non-stationary and volatile behavior along the time domain and depends greatly on system load. In this paper, a new hybrid forecast engine is proposed for SR requirement prediction. The proposed forecast engine has an iterative training mechanism composed of Levenberg-Marquadt (LM) learning algorithm and Real Coded Genetic Algorithm (RCGA). implemented on the Multi-Layer Perceptron (MLP) neural network. The proposed forecast methodology is examined by means of real data of Pennsylvania-New Jersey-Maryland (PJM) electricity market and the California ISO (CAISO) controlled grid. The obtained forecast results are presented and compared with those of the other SR forecast methods. (C) 2009 Elsevier Ltd. All rights reserved.