Service Oriented Architecture (SOA) is an architectural style that uses loosely coupled services, which have separate concerns, as the primary constituents to build software solutions as compositions of services. It emphasizes loosely coupling and interoperability of services, required for agility and flexibility of business processes (BPs), hence allowing current BPs as well as future BPs to be integrated. To design such SOA-compliant software that supports BPs, the service identification is a critical phase. Indeed, the form under which the service is located may impact the effectiveness of the SOA. More particularly, the granularity of the entire services is very crucial in attaining the flexibility. The main objective of this research is to mine BPs for SOA by introducing a new service identification approach. The approach first uses a BP mining discovery technique to discover the BP model. Then, it clusters the activities into tasks, and finally organizes tasks into an initial set of services. And checks the compliance of the resulted services against the service orientation principles. The proposed approach is validated by a case study. The approach helps large enterprises to mine their BPs, develop them, and identify services. Finally, we recommend an automation of the proposed service identification approach to making it more effective and useful BPs.