Unlike some natural catastrophes, there is no stopping a volcanic eruption. The best that can be done is to give people who live in the danger zones sufficient warning that they can escape disaster. Many volcanic areas are densely populated, largely because soils derived from lavas and ash are extremely fertile, and high volcanoes create decent rainfall because of their orographic effect. Naturally, nobody likes to up sticks, whatever the dangers, least of all if there are false alarms. As with seismic prediction, volcanologists do not have a good track record of foretelling big eruptions, even though a great many geologists cluster on and around volcanoes. Most of them flock to areas with active lavas, pyroclastic flows and other lugubrious after effects of major activity. However some do the painstaking work of trying to monitor the plumbing of volcanoes, to get a handle on which parameters are most likely to be authentic warnings of impending doom. It is no longer a matter of experienced volcano watchers and their instinctive feel for when one is about to blow its top, but one of ever more sophisticated instruments and software to analyse data and model volcanoes’ inner workings. The 28 March 2003 issue of Science (p. 2015-2030) devotes 16 pages to a review of volcano monitoring. While advances are being made, there is still a long way to go before they can pay dividends by reducing the loss of life. What is not going to go away, even in the best of all possible scientific worlds, is the economic devastation that follows any geohazard.