In B2B markets, when firms sign contracts for transactions pertaining to the exchange of services that are delivered over a period of time, one critical decision they make is the length (or duration) of the contract. If the services are hired for a long project, companies often sign multiple, successively run contracts with the same vendor. This is prevalent in projects such as when multinational companies hire consulting firms like Accenture to streamline and digitize their business processes, when big banks in developing countries hire firms like Tata Consultancy Services to extend banking facilities into rural markets, and when oil companies hire rig firms to drill oil wells. From a traditional economic perspective, companies would decide on an optimal contract length that is not too long or too short; the former disables the firms from reacting to market changes while the latter makes negotiation costs expensive. However, when a company signs a series of successive contracts with a service-firm, both companies get to learn about the other company’s goals and operations dynamically, which might influence the length of each contract in the series. Thus, determining the contract length in a series of successive contracts is more challenging. In this study, we build a contract length determination model that considers both the economic factors and the dynamic learning. The model provides managers with a theoretical yet practical tool to make optimal decisions on contract length. We use data from the oil-drilling industry to empirically test the proposed model.