This paper proposes a set of principles for assuring services (as opposed to systems) and examines how they may be applied in a service provision situation. These principles are intended to be generic and can be used across a wide range of service scenarios in diverse sectors. The principles are broadly based on the established software safety assurance principles but with a wider scope to cover all forms of undesired behaviour where assurance is required. The principles are linked to objectives and characteristics to show how high-level requirements can be established for assured services. Of particular significance is showing that the intent of the service assurance requirements is maintained through decomposition, i.e. through the service definition, service architecture and the agreements made at service interfaces, and that the principles are preserved through change. The approach draws upon concepts from modular assurance and describes the different types of “wrappers” that can be used to manage the variety of assurance positions presented by services: these can vary widely in nature, and the confidence in the assurance of intended behaviour can be difficult to establish. An example case study based on the BP Deepwater Horizon accident is used to illustrate the approach.