Grassroots fundraising leverages favor trading in social networks to support the provision of a public good. We use a laboratory experiment to study the elements and dynamics of this type of institution. Peer-to-peer reciprocity is important, and having the ability to practice targeted reciprocity in our experiment increases contributions to the public good by 14%. Subjects reward group members who have previously been generous to them and withhold rewards from ungenerous group members. When someone is rendered unable to benefit from favor trading, he gives much less to the public good than he does in other settings. People thus excluded from the “circle of giving” provide a clean and strict test of indirect reciprocity, since they cannot benefit from a norm of cooperation. Contrary to previous studies, we do not observe indirect reciprocity.