In the proposed frequency restoration plan, demand response is adopted as the first shedding option for intercepting frequency decline in order to avoid the unexpected load shedding, then followed by the scheduled generation reserve to raise frequency back to the normal state. This paper starts with the frequency response analysis using a low-order frequency response model. Results of the frequency response analysis show that, if the magnitude of system disturbance is accurately estimated following the moment of incident, the estimate could be intelligently used to deploy appropriate demand response for frequency restoration. Tests of the proposed frequency restoration scheme are evaluated by simulation where the system data is utilized by records of historical frequency events from a utility. Test results show that the deployment of the demand response could enhance frequency security under various contingency scenarios.