In regions where strong earthquakes occurred before the deployment of dense seismic and accelerometric networks, intensity datasets can help select appropriate ground motion prediction equations (GMPEs) for seismic hazard studies. This is the case for the Hainaut seismic zone, which was one of the most seismically active zones in and around Belgium during the twentieth century. A recent reassessment of the intensity dataset of the area showed that intensities in this region attenuate much faster with distance than in other parts of northwestern Europe. Unfortunately, this characteristic has not yet been taken into account in current hazard maps for Belgium and northern France. Based on this dataset, we evaluate the goodness of fit of published GMPEs with intensities in Hainaut by means of a ground-motion-to-intensity conversion equation (GMICE) and according to different metrics (Likelihood, Log-likelihood and Euclidean-based Distance Ranking) published in literature. We also introduce a new measure to specifically evaluate the distance trend. Our results show that none of the tested GMPEs convincingly fits the intensity dataset, in particular the fast attenuation with distance. Nevertheless, applying the few GMPEs that show a reasonable fit in seismic hazard computations, we observe a decrease of the influence of the Hainaut seismicity on hazard maps for Belgium and northern France. This result is compatible with the earthquake intensity observations for the last 350 years in this part of Europe.