Although earthquake prediction remains an ultimate goal of seismology, up to now only a very few events have actually been forecasted, while for many of them wrong or doubtful predictions were given and even some misunderstandings were produced. After reporting a couple of examples on the subject, the paper focuses on an accurate revision of the seismological data of the 1976 seismic sequence in Friuli (northeastern Italy), as well as on the analysis of the space distribution of the earthquakes aiming at investigating if any clue of possible identifi cation of the epicentre of the main aftershocks was possible at that time or should be possible today. The final consideration is that, even with a good (or good enough) seismic monitoring, there was no clear evidence of epicentre migration towards the future location of the major events.