Reputation mechanisms offer an efficient way of building the necessary level of trust in electronic markets. In the absence of independent verification authorities that can reveal the true outcome of a ransaction, market designers have to insure that it is in the best interest of the trading agents to report the behavior in transactions truthfully. As opposed to side-payment schemes that correlate a present report with future reports submitted about the same agent, we present a mechanism that discovers (in equilibrium) the true outcome of a transaction by analyzing the two reports coming from the agents involved in the exchange. For two long-run rational agents, we show that it is possible to design such a mechanism that makes cooperation a stable equilibrium.