A substantial amount of research has examined the collaborations within technology-based industries, but how technological intensity relates to the evolution of interregional research collaboration is less well understood. This paper investigates the dynamic changes of spatial structure and determinants of intercity research collaborations in the sectoral innovation networks of high-tech industries contrasting to low and medium technology (LMT) industries in the Greater Pearl River Delta (GPRD), China.Using the sectoral patent application data over the period 2008-2016, the social network analysis results confirm a higher level of triangles in the co-invention network of high-tech industries than within LMT industries. Moreover, the empirical results through Exponential Random Graph Model (ERGM) show that, compared to LMT industries, collaboration in the co-invention network of high-tech industries is less sensitive to geographical and linguistic distance, while more subject to economic distance, regional educational level and government support with time. In addition, cognitive proximity and regional R&D intensity are both important prerequisite in the formation of these two collaborative networks.