We present an automatic 3D city model of dense urban areas from high resolution satellite data. The proposed method is developed using a structural approach : we construct complex buildings by merging simple parametric models with rectangular ground footprint. To do so, an automatic building extraction method based on marked point processes is used to provide rectangular building footprints. A collection of 3D parametric models is defined in order to be fixed onto these building footprints. A Bayesian framework is then used : we search for the best configuration of models with respect to both a prior knowledge of models and their interactions, and a likelihood which fits the models to the Digital Elavation Model. A simulated annealing scheme allows to find the configuration which maximizes the posterior density of the Bayesian expression.