This research examines collective learning and the conditions for its occurrence in municipal collaborative networks. Collaborative governance, marked by horizontal coordination and shared goals, offers advantages for public policy, especially on cross-cutting issues. These include the development of more coherent, innovative, and legitimate policies through diverse perspectives. However, many governance processes fail to realize these benefits, leading to this study's focus on collective learning, defined as expanded understanding through repeated social interactions. Collective learning involves both individual learning (acquiring new knowledge) and mutual understanding (recognizing others' beliefs). The research uses comparative case studies of eight networks in four European cities, addressing issues like sustainable public procurement, climate change adaptation and anti-discrimination plans. Data collection includes interviews, mental models, and Social Network Analysis to capture learning and informal interactions. Our findings indicate that fostering collective learning is extraordinarily complex and requires the presence of several facilitating conditions. Firstly, diversity stands out as a necessary element with different forms of diversity contributing particularly to certain forms of learning. Secondly, interactive formal meetings are absolutely crucial for mutual understanding, as they allow for deliberative exchange of views. Finally, informal relationships indirectly facilitate collective learning by improving relational capital, which is particularly useful is siloed contexts. In addition, we observed that individual attitudes toward collaboration and learning significantly influence individual and thus collective learning. Typically, participants who present themselves as experts tend to report less individual learning.