Flooding poses a significant threat to infrastructure, ecosystems, and communities, exacerbated by climate change and urbanization. Effective flood governance requires coordinated action across sectors, yet current approaches remain fragmented and fail to address critical socio-environmental factors shaping collaboration. Despite increased awareness, the complexity of flood governance—driven by diverse stakeholders, varied risk perceptions, and uneven institutional capacities—remains a challenge. This study addresses these gaps by applying network analysis and Exponential Random Graph Models (ERGMs) to examine relationships among flood governance actors in the Beaumont-Port Arthur region. Our unique approach analyzes both the structural features of governance networks and the role of shared perceptions of flood risk drivers and impacts, offering insights into socio-environmental factors shaping collaboration. The paper identifies the structural characteristics of governance networks across flood preparation, mitigation, response, and recovery phases, while evaluating how perceptions of risk influence network ties. Our findings reveal significant governance gaps in addressing socio-economic and ecological impacts, with lower network connectivity in these areas. Conversely, infrastructure considerations play a central role in fostering partnerships, particularly in long-term planning efforts. Organizations with aligned perceptions of key drivers, such as infrastructure and precipitation risks, form more cohesive and responsive networks. These results emphasize the need for inclusive, adaptive governance frameworks that integrate diverse perspectives and strengthen local engagement. This study provides valuable insights for policymakers, offering pathways to enhance flood resilience by addressing the socio-environmental dynamics shaping collaborative networks.