Abstract: The physics engine technology is used to perform the seismic analysis of structures, considering both the damage scenarios of out-of-plane failure of infill walls and, of buildings collapse. The distribution law of debris is investigated according to the results obtained by the physics engine technology. this study clarifies the random distribution characteristics of post-earthquake building debris and points out the shortcomings of deterministic theoretical models. An unbiased probabilistic model for the debris extent is established with Bayesian updating rule. The posterior probability distribution of unknown parameters in the prediction model is developed, which provides an approach for quantifying the epistemic uncertainty associated with unknown model parameters. Taking a six-storey reinforced concrete frame (RCF) office building and a ten-storey RCF residential building as the research object, a probabilistic prediction model is presented for the width and area of post-earthquake building debris extent. The research results indicate that the maximum spacing of debris coverage has a huge variability, with a coefficient of variation (COV) ranging from 15% to 30%, while the COV of the area of debris extent is between 20% and 50%. The posterior probability density function curves of the unknown model parameters have multiple peaks. The traditional methods, namely taking a deterministic value for model parameters or describing the probability distribution of parameters with conventional distribution functions, will bring a considerable uncertainty.