The issue addressed in this paper is the objective selection of appropriate ground motion models for seismic hazard assessment in the Pyrenees. The method of Scherbaum et al. (2004a) is applied in order to rank eight published ground motion models relevant to intraplate or to low deformation rate contexts. This method is based on a transparent and data-driven process which quantifies the model fit and also measures how well the underlying model assumptions are met. The method is applied to 15 accelerometric records obtained in the Pyrenees for events of local magnitude between 4.8 and 5.1, corresponding to moment magnitudes ranging from 3.7 to 3.9. Only stations at rock sites are considered. A total of 720 spectral amplitudes are used to rank the selected ground motion models. Some control parameters of these models, such as magnitude and distance definitions, may vary from one model to the other. It is thus important to correct the selected models for their difference with respect to the magnitude and distance definitions used for the Pyrenean data. Our analysis shows that, with these corrections, some of the ground motion models successfully fit the data. These are the Lussou et al. (2001) and the Berge-Thierry et al. (2003) models. According to the selected ground motion models, a possible scenario of a magnitude 6 event is proposed; it predicts response spectra accelerations of 0.08–0.1 g at 1 Hz at a hypocentral distance of 10 km.