Abstract. Apologies are ubiquitous in life and across cultures and countries, and they are able to impact social relationships as well as material outcomes. One specific application refers to restorative justice and mediation programs. While empirical data is biased by unobservable influences, a controlled experiment provides additional evidence on the effect of apologies after exclusion. We present an experiment in which participants can steal from other group members and can be excluded for one period. We compare four treatments in which we give the option to apologize (or not) and where apologies can be tracked (or not). We find that apologizing significantly reduces theft and participants who apologize experience more re-integrative behavior from the rest of the group members than those who do not apologize. Additionally, we find that people do apologize even when their apologies cannot be tracked, suggesting that apologies are not purely used as a strategic device.