Scholars have long argued that in new product development integrated innovation processes replace sequentially organized ones because of changing capabilities, integration of knowledge, and the increasing importance of market orientation. The intention of this paper is to study whether this also applies in process-oriented industries. Process-oriented settings are fundamentally different from product-oriented ones as the emphasis is on efficiency, clear guidelines, and tacit knowledge, while radical innovation is less important. It is also less obvious why market orientation-a key driver of integration in new product development-should change the way the innovation process is organized, as there is no product to be developed in the first place. A Stage Gate Model could therefore be well-suited to achieve more efficiency and effectiveness. In order to investigate process innovation in a process-oriented setting, we decided to study the upstream oil and gas industry, a scale-intensive, process-oriented setting that substantially contrasts with traditional science-driven industries such as biotechnology and pharmaceuticals, where patents are relevant. The particular advantage of this setting is that virtually no product innovation occurs, allowing the isolation of process innovation. Relying on five inductive case studies of major projects (BP's Prudhoe Bay, Chevron's Kern River, Conoco Philips's Ekofisk, ENI's el-Bouri, and Shell's Troll field), we find that integration occurs in process-oriented settings but does so for different reasons than in product-oriented ones. Integration emerges from an innovation mode characterized by: (1) trial and error (not R&D) as the main mode of innovation; (2) the cooperation of experts from different knowledge backgrounds; (3) the development of Information and Communication Technology (ICT), which facilitates this cooperation across disciplines and projects; and (4) the need to increase efficiency as demand outstrips supply. More precisely, demand shapes reward systems and determines whether new raw materials will be used (creating new supply that requires innovation of processes in periods when demand outstrips supply). Supply and reward systems therefore create the conditions where trial and error, integration of knowledge, and ICT development mutually enforce each other, leading to the integration of the innovation process. Besides contributing to the literature on innovation the paper also offers interesting insights for managers. In order to foster innovation in process-oriented settings they need to provide specialists with room to experiment over extended periods of time, encourage cooperation across disciplines, and create both incentives and systems to facilitate this process. Finally, managers need to consider the ability of staff to cooperate at the outset, when they set up their recruitment process.