The performances of a mono-phase model and of a two-phase model have been evaluated carrying out a blind test. As a benchmark test the event that struck Sicily (2009) was chosen. Prediction accuracies have been evaluated determining statistical indicators and applying the ROC approach. For the simulated event the two-phase model is more accurate than the mono-phase one, though both models show limits when applied in a highly urbanized area, where some constrains cannot be properly reproduced.