How does a profit-maximizing manager form teams and compensate workers in the presence of both adverse selection and moral hazard? Under complete information, it is well known that any complementarity in characteristics implies that positive assortative matching is productively efficient. But, under asymmetric information, we uncover the problem of disassortative incentives: incentive costs may increase in assortativity. Profit maximization thus prescribes either random or negative assortative matching, both productively inefficient, when complementarities are weak and eort costs are high enough. When this is the case, the manager may instead prefer to dele-gate matching, allowing workers to sort themselves into teams. Our results shed light on recent empirical work documenting patterns of non-assortative matching inside of firms.