Pollution from nonpoint sources (NPS), and agriculture in particular, remains as one of the largest sources of water quality impairments in the United States. As is well known in the literature, there are many difficulties with designing regulations for reducing nonpoint source pollution (i.e., Tomasi, Segerson, and Braden, 1994). Uncertainty and asymmetric information are the key regulatory difficulties in the control of NPS. The main goal of this paper is to describe a potential incentive scheme that can be applied in limited information situations. The incentive scheme involves a contract written between a point source of pollution and a small group of other nonpoint polluters in the watershed to reduce a specific load of pollution. The contract allows the nonpoint sources to enter the contract voluntarily. To handle the incentive problems typical in many principal agent problems, it incorporates joint liability, and peer pressure/monitoring to induce the nonpoint sources of pollution to meet their contractual obligations. For this paper, we propose a group contract built upon the ideas of Stiglitz (1990) and Varian (1990), and originally applied to micro-lending arrangements in developing countries. As we hypothesize with nonpoint source pollution, joint liability contracts for micro-lending assume that individuals have more information about each other than the principal has about them, and they take advantage of joint liability and peer monitoring concepts to eliminate or reduce the moral hazard problem. Joint liability contracts have been shown to be successfully applied in practice in several situations (Ghatak and Guinnane, 1999; Van Tassel, 1999). The contract proposed for this paper assumes that a principal (a point source of pollution) offers a contract that specifies a price for each ton of pollution abated by individuals who participate. The contract is offered to individual farmers in a specified sub-watershed upstream from the discharge point. Farmers in the watershed decide whether or not to participate, and if they decide to participate, they bid into the contract the level of abatement services they will provide. The principal will form the group from these bidders, and will agree to pay the total amount if they meet the target. Given the fixed price for the contract, the bids determine the sharing rule for payments at the end of the season. With this basic idea, this paper explores the implications of moral hazard under several contractual arrangements. In particular, we are interested in the trade-offs between participation in the contract and shirking. At one extreme, existing voluntary incentive programs involve fixed payments with no requirements for performance. They will thus lead to high levels of participation, but likely to large levels of shirking as well. For example, in a typical year all of the money available for existing federal conservation programs is used by farmers, but there is little evidence available to prove that pollution declines as a result. At the other extreme, one could write a contract that specifies that payments will only be made if the target is met, following Holmstrom (1982) or Segerson (1988). Such an extreme, nonlinear contract such as this likely would eliminate most shirking, but it may lead to little participation among farmers. This paper thus explores the relationship between the participation constraint in principal agent problems and moral hazard with a simple theoretical model and numerical simulations. First, we propose a contract with a fixed initial payment and a bonus payment if the farmers meet the agreed upon target. The effects of the fixed payments upon the participation constraint and the consequent implications for shirking are shown theoretically and numerically. Fixed payments enhance the likelihood that individual farmers join the group, but fixed payments can also induce incentives to shirk. Second, we hypothesize that farmers can engage in other nonpecuniary penalties to punish farmers who shirk their responsibilities. For example, if someone in the group shirks, we assume that the group members can figure out whom, and take appropriate action to punish him or her. The numerical simulations show how the success of the proposed mechanism depends crucially on size of the fixed and bonus payments, group structure (i.e. ability and willingness to monitor each other) and group size. While not having any direct penalty in the contract, and having fixed payment plus bonus can increase the participation for the program, it can also increase free-riding problems in the group. Group size is also important, in that groups that are too large cannot take advantage of nonpecuniary penalties. Regions that have strong social ties are the most promising candidates for this kind of group contracts. They can be better able to apply social pressure on potential shirkers (Prescott, 1997). In addition to showing a theoretical model, and numerical simulations, we provide some results from recent focus groups providing data from actual farmers on their willingness to participate in the types of contracts proposed in this paper.