In this article, we tested two concepts of decision making: expected utility theory and heuristic choice. In Experiment 1, we applied think-aloud protocols to investigate violations of expected utility theory. In Experiments 2 to 4, we introduced a new process-tracing method—called predict-aloud protocols—that has advantages over previously suggested research methods. Results show the following: (i) people examine information between rather than within gambles; (ii) the priority heuristic emerges as the most frequently used strategy when problems are difficult; and (iii) people check for similarity when problems are easy. Copyright © 2012 John Wiley & Sons, Ltd.