The quality of numerical weather prediction has improved considerably since its beginning, however, this remarkable achievement has to be considered true for average conditions. It is known that atmospheric predictability and model errors are highly flow-dependent, therefore, an increase in skill for average conditions may not imply the same improvements in specific conditions. Moreover the potential value of numerical weather prediction is perceived to be higher in some specific cases, like high-impact weather events. There is therefore a growing need to know the forecasting accuracy of significant weather events, something that cannot be easily inferred through average scores, not least because of the rarity of these events. For these reasons, a study has been carried out to examine the skill of the European Centre for Medium-Range Weather Forecast (ECMWF) global forecasting system in predicting a specific flow configuration that is believed to be associated with extreme precipitation events over the Alpine region. Despite quantitative predictions of extreme precipitations is still challenging, it was found that the large-scale flow conducive to major rain events has better predictive skill than average conditions. This is perhaps surprising since it is a common perception to associate severe weather with low predictability.