Security-constrained optimal power flow (SCOPF) is a commonly used tool for power system planning and operation. Since a large number of contingencies are involved, it is very difficult to solve the problem. How to accelerate the solution process of the SCOPF has been a hot topic. Based on the traditional security-constrained optimal power flow according to screening of contingencies, this paper replaces the traditional severity index with overload risk and considers the probabilities of contingencies caused by real-time weather conditions. A new SCOPF algorithm based on overload risk screening of contingencies is proposed. Numerical results show that the proposed algorithm has a smaller contingency set, and is speedy, safely and economical. The proposed new iterative algorithm for the SCOPF can thus serve as a decision-making tool for the dispatchers.