are so much more catastrophic than the most likely scenario, one should not assume that the next event will be the most likely one.


Unlike most other known natural hazards to humanity, such as earthquakes, volcanic eruptions, tsunamis, hurricanes, and tornadoes, NEO impacts present a very large spread of disaster scales ranging from small property damage to global extinction events. Larger impacts may result in global climatic changes that can result in famine and disease, infrastructure failure and, potentially, societal breakdown. Smaller impacts could be misinterpreted and thereby could conceivably even trigger wars. Numerous small incidents present little risk to people and property, but major impact events occur very infrequently. Impacts represent the extreme example of “low-probability, high-consequence” events. Although the probability of such a major impact within the next century may be small, a statistical risk of such an impact remains. Because of the nature of the impact threat, the expected fatality rate from impacts is an “actuarial” estimate based on calculations with attempted conservative assumptions. All the other estimates in Table 2.2 are based on the attribution of causes of actual fatalities from ongoing threats that may change in the future.

In contrast to other known natural hazards, there has been no significant loss of human life to impacts in historical times, due to the low frequency of major impacts and the higher probability of impact in unpopulated

TABLE 2.2 Expected Fatalities per Year, Worldwide, from a Variety of Causes


Expected Deaths per Year

Shark attacksa








Traffic accidentse


Air pollutionf






NOTE: The entries in this table are of various types. For example, the fatality rates given for shark attacks, earthquakes, traffic accidents, and HIV/AIDS entries are extrapolations, based on past reported individual deaths due to these causes, estimates of the completeness of these reports, and the assumption that future such deaths will continue at the same average rates (or straightforward extrapolations from them). The asteroid impact entry has been treated in this chapter and is based on models for impact and tsunami effects, an assumption of ecological collapse above some global catastrophe threshold, and a statistical calculation of risk based on the known near-Earth-object size distribution, with the temporal rate expected to vary enormously from the rate given, that is, to be zero most years, sizable in a relatively few years, and enormous in only an extremely few years over a time span of a billion years. The entries for malaria and tobacco fatalities are inferences based on plausible assignments of causes of deaths; such assignments are, individually, far less reliable than, for example, is the case for shark attack fatalities. Mitigation Panel member Mark Boslough wanted an additional entry in this table for fatalities due to climate change. The Steering Committee disagreed with including this entry because it did not think that a reliable estimate is available, among other reasons. Dr. Boslough’s minority opinion is provided in Appendix D.

aData from International Shark Attack File,

bData from Harris (2009) and Figure 2.7.

cWorldwide, 1970-2009; data from U.S. Geological Survey, cited in

dData from

eData from

fData from and PHE_OEH_06.02_eng.pdf.

gData from

hData from

SOURCES: Data for this table were derived from the sources listed above, as well as the World Health Organization.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement