I will describe a series of studies done over the last dozen years via the Internet, investigating properties of risky decision making. The first was designed to test if properties we observed in the lab with students would also be descriptive of highly educated experts in the field of decision making. These people could be recruited via the Web and could participate via browser-based Web forms. As many have found, this method provides several conveniences, which led to its adoption in almost all studies done in my research program either with students in the lab or other people recruited via the WWW as well as tested that way. Use of Web studies has also been success in making the methods of experiments more transparent to others, in fostering cooperation between investigators at different institutions, and permitting quick tests of experimental hypotheses suggested by other researchers. Other useful practices are also supported via the Web; for example, one can archive data in a form that can be easily downloaded for reanalysis; one can archive calculation spreadsheets showing how to fit a theory to data; and one can place computer programs contained in Web pages, allowing users to make calculations using alternative theories. Such programs can be very useful to rivals planning to design new experiments to test between different models. The research program has found systematic violations of both versions of prospect theory. These violations are found with a variety of methods for displaying choices and representing probability, they are found in experts as well as among college students, and they are found in analysis of long experiments conducted with individuals.