Agents can benefit from contracting some of their tasks that cannot be performed by themselves or that can be performed more efficiently by other agents. Developing an agent’s contracting strategy in the University of Michigan Digital Library (UMDL), however, is not easy for the following reasons. The UMDL consists of self-interested agents who will perform a task of another agent’s only when doing it is their own interests. In addition, multiple contracts take place concurrently such that other contracts currently in the system may have an impact on the success of one’s own contract. Therefore, an agent who has a task (contractor) needs to model what the other self-interested agents think and will do, and it also needs to consider the influence of other contracts on its contract. In this paper, we define the contractor’s and the contractee’s decision problems in the UMDL contracting situations, and present a contracting strategy by which a contractor can determine an optimal payment to offer. The contractor’s problem is to find a paymenthat maximizes its expected utility, and it finds such a payment by modeling the contracting process stochastically using a Markov process. The Markov-process model is built based on the information the contractor has about the potential contractees and about the other contracts in the system. Our early results show that the contractor receives a higher utility when thinking about the potential contractees and the other contracts.