Abstract: Building on the co-publication data of inter-regional ‘industry-university’ scientific cooperation (2010-2015), this paper examines the influence of the structure of ‘industry-university’ scientific network on regional innovation performance at municipal level. Using negative binomial regression, the panel modeling reveals following findings. Firstly, the closeness centrality and clustering coefficient of network facilitate regional innovation, while the structural hole and city degree centrality shows ignorable effects on regional innovation. Secondly, the influence of network structural on regional innovation varies according to the extent to which the regional institution develops. In addition, the study corroborates these finding via several robustness checking. Lastly, this paper concentrates energy on discussing the related policy implications of these findings and avenue for further study.