In this experimental study of a two player power-to-take game, players earn an income in an individual effort task preceding the game. The game has two stages. First, one player can claim any part of the other's income (take rate). Then, the latter player can respond by destroying own income. We focus on how emotions in¯uence responses and show: (1) a higher take rate increases (decreases) intensity of negative (positive) emotions; (2) negative emotions drive destruction; (3) at high emotional intensity, responders destroy everything; (4) expectations affect the probability of punishment. Emotional hazard is identi®ed as a new source of ef®ciency costs.Many people would agree that emotions play a very important role in the decisions they make. Extensive research by psychologists over the last two decades has provided a lot of supportive evidence. It appears that emotions play a signi®cant role in matters like attention, learning, and memory (Izard et al., 1984). Recent neuroscienti®c research even suggests that emotions are essential for rational decision making (Damasio, 1994; Picard, 1997). If emotions play such an important role in psychological processes, they are also likely to be relevant for understanding economic decision making. Frank (1988) argues that emotions are relevant for economics because they can help us solve important commitment problems. He shows, for example, that players endowed with the emotion guilt can sustain the co-operative outcome of a prisoner's dilemma game. 1 Other recent economic studies focus on the effects of emotions on preferences (Hirshleifer, 1987;Loewenstein, 1996) or on the implications for rationality (Elster, 1996(Elster, , 1998. However, as yet little work has been done to integrate emotion theory in economic research. Elster (1998) addresses this neglect and hypothesises that it may have to do with the different explananda of psychology and economics:`Whereas economists mainly try to explain behaviour, emotion theorists try to explain emotions. By and large, psychological studies of the emotions have not focused on how emotions generate behaviour' (p. 47).The object of this study is to investigate how emotions generate behaviour in a laboratory experiment. As our vehicle of research we use a simple * This paper is part of the EU-TMR Research Network Endear (FMRX-CT98-0238). We are grateful to Joep Sonnemans, Theo Offerman, and Arno Riedl for discussions and comments, Jos Theelen for developing the laboratory software, and Jens Grosser for his assistance in the experiment. We are also grateful for the remarks made by participants of the ESA conference in Mannheim (1998), the IAAP conference in San Francisco (1998), and the ENDEAR Workshop in Bari (1999). Finally, we thank the editor David De Meza and three anonymous referees for their helpful comments.1 Frank (1988) assumes that people give signals about their emotional commitments or dispositions (for example, via facial expression or the pitch of the voice) that are dif®cult to simulate. In his prisoner's dilemma mo...