An immense amount of data is available immediately prior to, during, and in the direct aftermath of a disaster. This information is often referred to by hazards and disaster researchers as perishable data. Social scientists, engineers, and natural scientists have been collecting this type of data for decades, yet it is neither consistently defined nor discussed in specific detail in the literature. To address this gap in knowledge, this article seeks to clarify the meaning of perishable data and provide guidance on ways to improve how it is collected and shared. Specifically, we review available definitions and offer an expanded conceptualization of perishable data as highly transient data that may degrade in quality, be irrevocably altered, or be permanently lost if not collected soon after it is generated. In this revised definition, perishable data may include ephemeral information that must be collected before, during, or after a disaster to characterize pre-existing hazardous conditions, near-miss events or actual disasters, and longer-term recovery processes. This data may need to be gathered at multiple points in time across varying geographic scales to more accurately characterize exposure, susceptibility to harm, and coping capacity. The article examines various ethical and logistical challenges for collecting perishable data in different cultural contexts. The article concludes with a discussion of opportunities to advance this type of data collection and its dissemination while emphasizing the role that perishable data collection can play in the evolution of the hazards and disaster field.