Purpose – The purpose of this paper is to recover the narratives constructed by the disaster management policy network in Washington, DC, about the management of Hurricanes Katrina and Sandy. Recovering and analysing these narratives provides an opportunity to understand the stories constructed about these events and consider the implications of this framing for post-event learning and adaptation of government policy. Design/methodology/approach – This research was conducted through an extended ethnographic study in Washington, DC, that incorporated field observation, qualitative interviews and desktop research. Findings – The meta-narratives recovered through this research point to a collective tendency to fit the experiences of Hurricane Katrina and Sandy into a neatly constructed redemption arc. This narrative framing poses significant risk to policy learning and highlights the importance of exploring counter-narratives as part of the policy analysis process. Research limitations/implications – The narratives in this paper reflect the stories and beliefs of the participants interviewed. As such, it is inherently subjective and should not be generalised. Nonetheless, it is illustrative of how narrative framing can obscure important learnings from disasters. Originality/value – The paper represents a valuable addition to the field of disaster management policy analysis. It extends the tools of narrative analysis and administrative ethnography into the disaster management policy domain and demonstrates how these techniques can be used to analyse complex historical events.