We provide empirical evidence that an early form of “mobile money” is used to transmit funds to individuals a?ected by catastrophic shocks. Contrasting two stylized models of prosocial behavior, we further provide insight into why people help each other in times of dire need. Our ?ndings are based on the analysis of billions of mobile phone-based transactions that occur before and after a destructive earthquake in Rwanda. The observed pattern of transfers is not consistent with a model of pure charity or altruism, but better ?ts a model of instrumental reciprocity. This conclusion is supported by three distinct results. First, earthquake-induced transfers are increasing in the wealth of the recipient, and are not signi?cantly related to the wealth of the sender. Second, transfers sent in response to the earthquake are highly dependent on the prior history of transfers. Third, transfers decrease as the distance between sender and recipient increases, even after controlling for the strength of pairwise relationships. Taken together, the evidence indicates that Rwandans use the mobile phone network to help a?icted friends and family, but that these gifts are motivated, at least in part, by a desire for reciprocity