Progressive collapse is commonly defined as large-scale catastrophic structural failure that is triggered by local structural damages, which usually leads to significant financial losses and human casualties. Understanding the risk of progressive collapse of buildings is of paramount importance for the design of the necessary protective measures. This paper presents a stochastic numerical model, which facilitates an efficient computation of the risk of progressive collapse of reinforced concrete (RC) buildings. In the proposed model, all the structural components such as beams, columns and beam-column joints are modeled as several elastic or rigid blocks connected by nonlinear cohesive elements, which represent the potential damage zones. The constitutive behavior of the cohesive elements can be determined by fine-scale finite element (FE) simulations of the potential damage zones under various loading conditions, such as uniaxial tension, compression, pure shear, and also mixed-mode loading. The cohesive elements are deleted once they lose their load carrying capacity, and the corresponding structural component becomes disintegrated. The fracture energies of the cohesive elements are then randomized to investigate the probabilistic collapse behavior of a two-dimensional 30 storyRCstructural frame under different local column removal scenarios, and the corresponding risks of various possible collapse extents are quantified.