INTRODUCTION The Pearl River Delta in Southern China (PRD) is a collective name covering the triangular region around the confluence of the Pearl River system that stretches across Guangdong Province of Southern China. The relocation of manufacturing plants from Hong Kong to Guangdong Province began in the early 1980s (Information Service Department, 2002) and is now virtually complete, leaving Hong Kong with only the related manufacturing services. The PRD region had more than 53,000 of manufacturing establishments in 2001 (Federation of Hong Kong Industries, 2002), while the total export value for the region in the same year accounted for 4.7% of the world merchandise trade (Federation of Hong Kong Industries, 2002). The PRD has thus developed into a very important member of the global manufacturing establishment (Zhao, Lo, & Sculli, 2005). The flow of manufactured items within a company and between companies has now become an important area of research, and at a macro level, these companies play their individual roles along the entire value chain. The structure and nature of the underlying industrial networking strategies has, to the authors' best knowledge, received limited attention from researchers, and a logically defined classification of industrial network strategies is needed in order to obtain a deeper insight into industrial networking. The results presented in this paper, are possibly a first attempt to devise a framework for the development of the associated industrial taxonomy using empirical evidence collected from manufacturing firms in Southern China. The two main industrial zones of Southern China are Hong Kong and PRD, and the taxonomy developed applies to the industrial networks of the manufacturing industries across both zones. Classification in a very broad sense is the grouping of similar objects (Everitt, 1993). It is a rather primitive method for categorizing objects into groups that are characterized by similarities and differences that the other groups do or do not possess to the same extent. Gordon (1996) suggests that the classes are discovered through the process of classification and are unknown a priori. He defines classification as: ... the investigation of a set of objects in order to establish whether or not they fall naturally into groups (or classes, or clusters) of objects with the property that objects in the same group are similar to one another and different from objects in other groups; these groups are unknown at the start of the investigation, and need to be determined. In recent years, cluster analysis has been the main technique used to reveal the underlying structure of various empirically based Operations Management (OM) topics, including advanced manufacturing technology (AMT), quality management system (QMS) and supply networks. Diaz, Machuca and Alvarez-Gil (2003) applied hierarchical cluster analysis to determine the appropriate number of groups for AMT investment patterns in the aeronautical industry. They identified 3 relatively distinct groups: traditionalists, designers, and investors who represent the underlying structure of the industry. Yeung, Chan and Lee (2003) adopted the hierarchical procedures to identify 4 specific groups of electronics manufacturing firms that are significantly different in their practice of QMS. Harland, Lamming, Zheng, and Johnsen (2001) employed cluster analysis to define 4 types of manufacturing firm that have distinct characteristics in supply networking strategies. Evidence of the growing popularity and acceptability of the use of cluster analysis in OM research can readily be found (Miller & Roth, 1994). In order to address network classification in industrial networks, Rudberg and Olhager (2003) suggest two distinct research tracks: supply chain research and manufacturing networks. For the first research track, i.e. research on supply chain, Fisher (1997) proposed two distinct types of supply chain according to the demand nature of the products--functional or innovative products. …