Prediction of natural phenomena which have the potential for causing disasters is an extremely difficult proposition both from a scientific and socio-political perspective. Experience has shown repeatedly our inability to predict events leading to disasters. Geophysical predictions, in order to have maximum utility, must specify the date, time, place and magnitude of physical events. In order to meet the minimum criteria for effectiveness and credibility, predictions must be stated within limits which are useful to the public and can be practically applied. Perhaps the best way to deal with the inevitable conflicts and uncertainties associated with the problem of disaster prediction is to either not make such pronouncements or issue forecasts based on the likelihood or probability of event occurrence. In the latter case it is best to provide simplified thresholds for various threat levels and suggest appropriate actions necessary to avert the impacts of an event.