A cascading outage model based on dynamic simulation is developed and analyzed in this work together with its stochastic counterparts， such as stochastic characters of generator frequency/voltage protection， under-frequency/under-voltage protection， over-excitation， overload， short circuit and the tripping time. The model captures the fundamental dynamic characteristics of power systems and is valid for studying system cascading outage. Based on PSS/E platform and Monte Carlo method， a test on practical system is given which shows the curve of outage scale versus outage frequency obeys the exponential distribution and has critical points. When the critical point is exceeded， the outage size and its probability rise rapidly and the power law is changed. Detection of critical points can provide helpful information for blackout prevention.