The competitiveness cluster policy (CCP hereafter) has been launched in 2005 in France to catalyze innovation:financial support is provided to collaborative R&D projects involving at least one public actor and two firms(whatever their size), thus encouraging the inter-connection of innovative actors. According to the CCP latestevaluation report, SMEs remain underrepresented in the network of innovation. In such a context, the presentpaper ambitions to investigate SMEs networking patterns for innovation, their determinants and peculiarity withinthe overall innovative network supported by CCP. SMEs sound disabled in running collaborative innovation, due totheir limited resources and competences in establishing/maintaining networks of collaborations (Kaufmann andTödtling, 2002) and their reduced absorptive capacities (Spithoven et al., 2010). To overcome this problem, SMEsmight benefit from the support of Knowledge Intensive Business Services (KIBS) (Shearmur and Doloreux, 2009;Muller and Zenker, 2001). But those KIBS are mostly SMEs themselves (Strambach, 2008). Studying openinnovation practices in the very context of SMEs, Van de Vrande et al. (2010) also exhibit differences betweenmanufacturing and services firms on the one hand, and between medium-sized and small firms on the other. Allthose findings call for a more fine-grained analysis of SMEs’ involvement in innovation networks. The diversity ofcollaborations for innovation concretely managed by SMEs also remains unclear. If for Rothwell and Dodgson(1994) SMEs collaborations tend to be limited to strategic alliances with large firms, the homophily argument(Ahuja et al., 2009) suggests that collaborations among SMEs would be privileged. The question of SMEscooperation with public research is also relatively inconclusive. Finally, we not only characterize SMEs likelihood totake actively part to the innovation network, but also study their respective positions in the innovation network,and precise the diversity of their collaborative behaviors. To do so, we run a network analysis based on 549collaborative projects funded by the financial instrument (FUI) associated to the CCP. In a first step, we work onthe overall network of interactions. Then, following Autant-Bernard et al. (2007) and Balland (2012), we eliminatepunctual participations to FUI projects. We thus distinguish “marginal” innovative collaborators (partners with asingle collaboration among one another), vs “key” ones (partners recurrently collaborating in FUI projects), andanalyze size influence on firm likelihood to belong to each population. Third, we identify innovative communities inthe network constituted by “key” partners and study the presence/ position of SMEs in those communities, and thedeterminants of the various collaborative behaviors observed (by distinguishing between different types of SMEs).