While social network analysis continues to enjoy considerable attention, literature on social network data collection often lacks explicit attention to methods. This presents special challenges to approaching the problems of undertaking social network analysis and of studying disaster preparedness, planning, and, ultimately, risk reduction. In this paper, we address this issue by presenting our synthesis of several strategies for network analyses from our processes for network identification and data collection in a longitudinal study of multi-jurisdictional, inter-agency wildfire response networks in the American Northwest. In the course of this ongoing project, the process of detecting and collecting data on pre-existing and emergent networks in the real world was not a matter of one theoretical or empirical judgement, but rather several. We alternated between: (1) spatio-ecological detection of jurisdictions adjacent to areas at-risk for large wildfires; (2) a hybrid approach to selecting actors and agencies identified as common participants in wildfire response networks; and (3) event-based detections of parties to specific wildfire response networks. We conclude with steps for thinking through network identification and bounding, integrating networks, conceptualizing rosters and ties in initial and events-based phases, and how to manage longitudinal network data collection.