Abstract: State Key Laboratories (SKLs) include labs affiliated with universities and research institutes (hereinafter called ‘Academic State Key Laboratories’, ASKLs), and labs affiliated with enterprises (hereinafter called ‘Enterprise State Key Laboratories’, ESKLs). SKLs are important drivers of China’s scientific and technological innovation. This study investigates the relationship between the number of “top design” projects and “free exploration” projects undertaken by SKLs and their impact on the probability of achieving scientific breakthroughs. Additionally, it examines the moderating effects of the structure of disciplinary knowledge networks on this relationship. The results of this study provide a comprehensive understanding of how project types and organizational characteristics simultaneously influence scientific breakthroughs. The empirical analysis was based on a dataset that came from annual laboratory reports and publication data from 185 SKLs spanning from 2010 to 2022. The dataset included 108 ASKLs and 77 ESKLs. The publication data was retrieved from the Web of Science Core Collection, which included a total of 87,329 papers published by these SKLs, including 8,047 highly cited papers. The establishment of disciplinary knowledge networks was based on the co-occurrence relationships among the disciplines indicated by SKL publications. Ucinet software was used to create a dynamic network with a rolling three-year window, resulting in 1,499 disciplinary knowledge networks. Based on the above, we obtained a non-balanced panel dataset covering the years 2010 to 2022. Our findings revealed that the number of “top design” projects was positively correlated with the probability of achieving research breakthroughs. In contrast, the number of “free exploration” projects had no statistically significant effect on breakthrough probabilities. This suggests that SKLs have more pronounced advantages in scientific research in the context of “top design” projects. Furthermore, the relationship between the number of “top design” projects and their breakthrough probabilities was moderated by two characteristics of disciplinary knowledge networks: betweenness centralization and cluster coefficient. These insights highlight that SKLs with strong interdisciplinary research backgrounds and deep engagement in specific fields are better able to leverage their “top design” projects to achieve scientific advancements. However, the influence of disciplinary knowledge networks on the breakthrough probability of “free exploration” projects is significant only for ASKLs, while no such effect is observed for ESKLs. This indicates that ESKLs' innovative capacities, particularly for “free exploration” projects, may rely more on the innovative capabilities of individual researchers than on their existing knowledge base. The academic contributions of this study include: First, it is the first time that the influence of the number of "top design" projects and "free exploration" projects on the probability of scientific research breakthroughs has been empirically demonstrated, which enhances the understanding of the advantages of national scientific research breakthroughs. Second, it provides the first empirical evidence that scientific research breakthroughs depend not only on the knowledge base but also on the types of research projects. Third, it shows that the contribution of the knowledge base to scientific breakthroughs is related to the type of laboratories. For "free exploration" projects, the moderating effect of the disciplinary knowledge network exists only for ASKLs, not for ESKLs. This means that it is difficult for Chinese enterprises to improve the probability of "free exploration" scientific research breakthroughs by relying solely on their own knowledge base. The practical suggestions of this study include the following: First, the evaluation of "top design" science and technology projects should consider the characteristics of the applicant's disciplinary knowledge network (such as betweenness centrality and clustering coefficient), and focus on its disciplinary accumulation and cross-disciplinary application experience. Second, labs with rich disciplinary knowledge bases should be encouraged to apply for "free exploration" projects, but the names of the labs should be anonymized to avoid potential biases from reviewers based on lab prestige. Meanwhile, the rules of the project application process should allow applicants to fully present their high-quality academic achievements. Third, national innovation strategies should emphasize the cultivation of a conducive knowledge environment, facilitate the formation of disciplinary knowledge networks, and promote interdisciplinary research through flexible talent policies and cross-disciplinary project funding.