Recent, unprecedented increases in wind penetration pose new challenges to maintaining good control performance in power systems. The stochastic nature of wind power creates an incentive to treat control resources as stochastic. This paper presents a stochastic, time-dependent model of contingency reserve (CR) activation that is based on system behavior expected by the NERC and WECC disturbance control performance standards. The model quantifies CR activation as a statistical combination of probability density functions (PDFs) that represent CR response deficiencies and generation outage probabilities. This paper also formulates an algorithm for determining the CR response deficiency PDFs in real time from observed energy management system data. The proposed model has multiple applications such as a deterministic or stochastic unit commitment, system adequacy evaluation, contingency reserve margins, and contingency reserve allocation.