We investigate two previously little studied aspects of favor-trading. First, we study whether and how individuals who stand to gain from favor-trading can best form cooperative relationships in an environment with private information about each agent’s ability and willingness to do favors. For agents with a low discount factor (low types) cooperation is not incentive compatible, for agents with a high discount factor (high types) it is. Both types receive privately observed opportunities to do favors with positive probability each period. We show high types are always able to separate from low types. Separation is implementable as soon as a high type receives a favor opportunity if the opportunities are independent across agents. If they are mutually exclusive, high types continue to separate with probability one if one of the agents is designated to do the first favor and that agent is a high type. Equilibria that designate an agent to act first implement separation approximately twice as slowly as symmetric equilibria that task the first high type with opportunity to separate first. Therefore the latter type of symmetric equilibria may dominate the former type of non-symmetric equilibria. Second, we study two-player games of favor-trading in a complete information environment standard to the literature, but in contrast to prominent models of favor-trading to date, we assume agents have concave utility functions of the form u(x)=x^a; 0