Contingency screeningand ranking is an important component of probabilistic security assessment of power system. In order to make the security assessment be monitored real-timely and applied online, a new model of contingency screening and ranking considering both accuracy and real time is proposed. To meet the requirements of both accuracy and speed, the model uses improved small-range analytic method to calculate small-range dynamic security region which is sensitive relative to the accuracy of probabilistic insecurity index. The stability criterion is used to classify the faults, and various faults are ranked using different strategies. Fatal faults are ranked according to probabilistic insecurity index, and then output immediately. For general faults, it defines the most easily instability distance based on small-rangedynamic security region, optimizes fault loss threat model, preliminarily screens the faults according to instability distance, and then ranks them based on loss threat. The new proposed model calculates each fault’s dynamic security region once, ranks fatal faults once, and screens and ranks general faults once. Simulation on IEEE 10-generator 39-bus system shows that the new model can make contingency screen and rank quickly and reliably, and the results have strong guidance for operation schemes of power system. This work is supported by Science and Technology Project of Northeast Branch of State Grid Corporation of China (No. DBDW2016-002).