Increased attention has been drawn to the phenomenon of community-based innovation; however, researchers have mainly emphasized firm-centric communities, paying little attention to non-firm members. This study focuses on university–industry (U–I) innovation communities to address this gap. Using China’s joint patent application data from 2000 to 2017, we construct a U–I collaboration network, identified existing U–I innovation communities, and reveal community-level characteristics of birth, expansion, maturity, and self-renewal in a lifecycle framework. The results suggest that average geographical distance negatively affects firm invention production and within-community knowledge diversity positively affects firm innovation. The effects of dynamic attributes indicate that U–I community membership turnover affects member firm’s patent production in an inverse U-shaped manner. In addition, a firm’s within-community network position exerts a moderating effect on the relation between community membership dynamics and firm innovation. In theoretical terms, this study combines innovation community and social network theories, using a lifecycle framework to examine the influence of the specified characteristics in facilitating member firms’ invention productivity. Finally, this study discusses the practical implications for U–I community stakeholders and policymakers.