While the number of artificial space object re-enter the Earth atmosphere daily, predicting the reentry of a space debris remains an open problem. The reentry prediction is a multi-physics problem involving aerodynamics computations in rarefied and continuum flow, heat transfer calculations and structiral breakup predictions. Additionally, numerous uncertainties coming from unknown initial flight conditions, material properties or uncalibrated model parameters affect oru ability to make acurate predictions. In this wrk, we propose an original reentry prediction framework that associates deterministic physical solvers with a stochastic breakup model and uncertainty quantification tools to make robust reentry predictions and a statistical estimate of the impact location. Our method is able to predic breakup distributions and ground impact locations efficiently using simplified but robust models at reasonable computational cost. This framework is used to predict the reenty of Upper Stage deorbited from a GTO orbit.