This paper attempts to develop concepts of project and contract organization to predict the selection of contract type on infrastructure projects. Conventional wisdom is that at low-risk fixed price contracts are best, moving to remeasurement and then cost plus as risk increases. We started trying to predict this from a transaction cost perspective, and such an analysis confirmed conventional wisdom. However, it does not fit with current practice. Further, the differences in transaction costs are small costs are small compared to differences in contract out-turn cost that occur under the different motivational effects of different contract types. We therefore take a different perspective. We assume the purpose behind a project contract is to create a cooperative project organization, in which all participants, clients and contractors, are motivated to achieve common objectives, their goals are aligned. This analysis confirms modern practice, and shows selection of contract type is related to uncertainty in the project's deliverables, and uncertainty in the process of their delivery. Build-only remeasurement contracts are used where uncertainty of both product and process is low. Designed and build fixed price contracts are used where uncertainty of the product is low, but the uncertainty in the process of delivery is high. Fixed contracts should be used where both are high. We extend the analysis to show when the client should be involved in the project organization in an alliance contract, and when they should not, as in a traditional project contract.