Friends often help each other without direct compensation. This behavior seems to contravene a basic tenet of economics, that prices and payment help coordinate supply and demand, increasing welfare. If compensation is useful between strangers, why not between friends as well? I address this question by building a model of bilateral trade in a friendship, where the price must be specified ex ante, and use it to understand under which circumstances a price of zero is an equilibrium outcome. In the model, two people have altruistic preferences towards each other, but may also have incomplete information about each other's motives. While altruists are able to undertake some beneficial transactions even at a price of zero, they would still benefit from a more efficient price. In contrast to previous work, I find that choosing to forgo payment cannot arise from being altruistic, nor from wanting to signal one's altruism. The key driver of this result is that the benefit of an altruistic reputation is greater to a selfish person than to an altruistic one, so no separating equilibrium in a signaling game could be sustained in which only the altruistic type sets price to zero. Instead, I find that a separating equilibrium is possible where a price of zero signals a partner’s _trust_ in the other’s altruism, though this result can only be sustained at extreme parameters. Overall, my model rules out altruism signaling as a fundamental reason for the lack of priced transactions between friends, but opens the possibility that this behavior could be driven by signaling of trust. I conclude by discussing alternative approaches to examining this widespread behavior.