Several blackouts and large-area outages around the globe were caused by cascading failures in the past several years. The ability to forecast cascading failures has become one of the most important research topics in power systems.The existing methods of forecasting cascading failures do not systematically and comprehensively address some important influencing factors such as the network topology，the relevance between two consecutive failures，and the uncertainties in the system operations. Given this background，based on the initial failure probabilities of lines and the transition probability between two consecutive failures，a Markov chain based model for forecasting cascading failures is presented.In this model，the transferring power flow，malfunctions of protective relays and circuit breakers，and the failure rates of components are considered.At first，the next failure is forecasted by the Markov chain based method，and the set of possible failures is identified，followed by the paths of cascading failures with the time sequence determined.Then，three criteria used to evaluate the vulnerability of follow-up failures are presented，and employed to monitor and quickly control possible cascading failures.Finally，the IEEE10-unit39- bus system is used for demonstrating the essential features of the developed model and method.