Wes Longhofer came a cross a new database: The Centre for Research on the Epidemiology of Disaster’s International Disaster Database. Site: http://www.emdat.be/
The site has cross-national data on both natural and human-caused disasters since 1900. Apparently, the most costly industrial accident in history was a chemical spill in Spain in 2002. Didn’t know that…
The dataset will be useful for our papers on environmental associations/policy reform/etc. Our prior work has generally found that environmental degradation variables (e.g., pollution) don’t do a good job of accounting for environmental mobilization or policy reform. Reviewers have then suggested, on more than one occasion, that people may respond to vivid disasters (Three Mile Island, Exxon Valdez, etc), rather than actual degradation. So, at one time David Frank pulled together a simple measure of disasters… but now someone has assembled a much more systematic dataset.
Disasters might also be an interesting issue to analyze as a dependent variable. For instance, one wonders if strong environmental/health/safety laws, strong unions, or other factors reduce industrial accidents… Maybe INGOs help, too… they do everything. (kidding…)
World Society, Institutional Theory, and Globalization 
It probably isn’t a new database, but it is new to me. INGOs don’t do much to prevent natural disasters (if they did, they would need a better superhero name), but they may mitigate deaths following some disasters. (Probably not, I know.) This database is being used in Michael Barnett’s work on humanitarianism, so I imagine it is something that can be used for a lot of different kinds of world polity research.
i think timmons roberts is also using these data for a big project.
Thanks for the tip, Andrew. See you in Vegas?