The costs associated with natural disasters are increasing rapidly. As a result, officials in government and industry have focused more attention on disasters and their effects. The White House Office of Science and Technology Policy has estimated that disasters cost the country about $1 billion per week. Hurricane Andrew, the Midwest flood of 1993, and the Hanshin earthquake have shown that individual disasters can cost tens if not hundreds of billions of dollars. This increasing cost has resulted in greater funding from government and industry for the development of technologies related to disaster prediction, and has led to more research into the effective use of predictive information. The insurance industry has long been aware of the dangers of natural disasters the 1906 earthquake in San Francisco, California, bankrupted scores of insurance companies. But the industry has focused particular attention on disaster prediction in recent years, as spiraling costs revealed that many companies had underestimated their financial exposure. For instance, prior to Hurricane Andrew in 1992, many insurance experts thought that the worst hurricane possible would do no more than $8 billion in damages to the industry. The damages caused by Hurricane Andrew, estimated at about $17 billion, shattered these beliefs. Today, estimates of worst-case disaster scenarios approach $100 billion. The insurance industry has therefore increased its support for research into disaster prediction. One such effort involves a number of companies that have joined together to support the Bermuda-based Risk Prediction Initiative, which funds disaster research. The expectation is that the resulting information will place the industry on a more solid foundation to make decisions about the risk of future disasters. The industry has also lobbied for the government to bear some of the financial burden of disaster insurance. Such a programme already exists for flood insurance, set up in the late 1960s by the federal government to insure flood-prone areas. These types of programmes, effectively implemented, could be increasingly necessary in the future to make insurance available in areas prone to disasters. Because the stakes are so high, the science of disaster prediction has a bright future. The various projects and programmes illustrate that disaster prediction is a topic of concern to scientists and policy makers alike. Hurricanes, tornadoes, floods, earthquakes, tsunamis, and volcanoes all show that the effective use of disaster predictions not only requires advanced technology but also requires that society consider the entire process of prediction forecasts, communication, and use of information. Because they cannot predict the future with certainty, and because much remains to be learned, scientists warn that society must understand the limits of scientific predictions and be prepared to employ alternatives. Wisely used, however, disaster prediction has the potential to reduce society's vulnerability to natural disasters. The result of the increasing costs in natural disasters is