We use data from a television game show involving elementary lotteries as a natural experiment to measure risk attitudes. Unique features of our data set are the substantial monetary stakes and the large sample size. CRRA and CARA utility specifications perform approximately equally well. We find robust evidence of substantial risk aversion. Extensions of the basic model, which allow for a separate utility flow purely from playing the game or for decisions based on decision weights instead of actual probabilities, raise the estimated degree of risk aversion.