This paper investigates behavior in the Traveler's Dilemma game and isolates deviations from textbook predictions caused by differences in welfare perceptions and strategic miscalculations. It presents the results of an experimental analysis based on a 2x2 design where the own and the other subject's bonus-penalty parameters are changed independently. We find that the change in own bonus-penalty alone entirely explains the effect on claims of a simultaneous change in one's own and the other's bonus-penalty. An increase in the other subject's bonus-penalty has a significant negative effect on claims when the own bonus-penalty is low, whereas it does not have a significant effect when the own bonus-penalty is high. We also find that expected claims are inconsistent with actual claims in the asymmetric treatments. Focusing on reported strategies, we document substantial heterogeneity and show that changes in choices across treatments are largely explained by risk aversion.