This paper extends previous models of rail freight system capacity and capacity flexibility in three important ways. First, uncertainty in the future traffic pattern is included; second, volume-delay functions and a level-of-service constraint are added to represent deterioration of service quality as capacity limits in individual facilities are approached; and third, a stochastic traffic assignment procedure is added to eliminate the need for path enumeration and to make the model more useful for large networks. These enhancements make the model suitable for use in the assessment of the performance of a freight network under conditions in which individual links or terminals have degraded capacity. An illustration of the model as applied to the intermodal double-stack container network in the western United States indicates the improved estimates of capacity and capacity flexibility that the model provides.