As envisioned by Intelligent Manufacturing Systems (IMS), Next Generation Manufacturing Systems (NGMS) will satisfy the needs of an increasingly fast-paced and demanding market by dynamically integrating systems from inside and outside the manufacturing firm itself into a so-called extended enterprise. However, organizing these systems to ensure the maximum flexibility and interoperability with those from other organizations is difficult. Additionally, a defect in the system would have a great impact: it would affect not only its owner, but also its partners. For these reasons, we argue that a service-oriented architecture (SOA) would be a good candidate. It should be designed following a methodology where services play a central role, instead of being an implementation detail. In order for the architecture to be reliable enough as a whole, the methodology will need to help find errors before they arise in a production environment. In this paper we propose using SOA-specific testing techniques, compare some of the existing methodologies and outline several extensions upon one of them to integrate testing techniques.