Diagnosis is an important function of a holonic manufacturing system if the desired levels of stability, adaptability and flexibility are to be achieved. Our research agenda is to study holonic behaviours (such as diagnosis and control) through the incorporation of these behaviours into operational industrial systems. Given the lack of fielded holonic solutions in industry, we are currently constrained to use conventional systems in our work. In this paper we describe the development of a holonic diagnostic capability for a PLC-controlled vehicle assembly line. A novel model-based strategy is used for diagnosis. Because of the constraints imposed on model formation in this environment, a two-phase approach consisting of off-line fault space generation and online fault space analysis is used. The fault space analysis utilises heuristics to achieve the desired performance levels (diagnosis in less than 60 seconds and success rates of greater than 90%). Areas for further research in holonic diagnosis are identified.