The workflow orchestration and traceability (or provenance) of data is essential in carrying out meaningful analyses of large scale neuroimaging datasets. In both the neuGRID and N4U projects the need for data provenance has been identified by the end-user research community as an essential requirement. We have addressed this through the provision of a Provenance Service and workflow orchestration whose description is the main contribution of this article. In the neuGRID project, a provenance service has been designed and implemented (using the CRISTAL software) that captures, stores, retrieves and reconstructs the workflow information needed to facilitate users in conducting neuroimaging analyses. The software enables neuroscientists to track the evolution of workflows (or pipelines) and datasets. It tracks the outcomes of various analyses and provides provenance traceability throughout the lifecycle of their studies thereby supporting the complex analyses underpinning neuroimaging research. Finally, the neuGRID Provenance Service and infrastructure has been evaluated in practice with clinical users and via a number of international data challenges.