We report a test of the potential of satellite data for the detection and forecasting of hail. It was found that the forecasting time from the first appearance in visible Meteosat data to the occurrence of hail is about 90 min. A necessary but not sufficient condition for the occurrence of hail is, that a given threshold in visible Meteosat data must be exceeded. Predictions based on this visible threshold in Meteosat data yield a probability of detection (POD) of 100% and a false alarm rate (FAR) of 81%. Additional meteorological measurements like sounding data reduced this FAR below 60%. Furthermore, it was found that almost half of the hail events produced hail when the storm was in its growing stage. Before the onset of hail either an existing cluster doubled its size or a new one appeared. This feature seems very promising for short-range forecasting and nowcasting purposes.