The rapid increase in installed wind generation capacity in the United States has raised concerns about electricity system reliability because of the intermittent and variable nature of the wind power. To help manage the wind power variability new ways of determining the amount of operating reserve capacity that must be kept available are necessary. A two-stage stochastic programming approach for the unit commitment problem had been introduced to mimic the short-term power operation decision process, i.e., day-ahead unit commitment and hour-ahead economic dispatch. Within this framework, a new formulation is proposed to determine the spinning and non-spinning reserve levels for large-scale systems over a 24-h optimization horizon with economic considerations. The proposed model is then applied to a large-scaled California test system. Simulation results illustrate the impact of stochastic wind power generation behavior and increased installed wind capacity on operating reserve requirements and the system cost as a whole. (C) 2011 Elsevier Ltd. All rights reserved.