OBJECTIVE: To document and explore the views and experiences of key stakeholders regarding the consent procedures of an emergency research clinical trial examining immediate fluid resuscitation strategies, and to discuss the implications for similar trials in future. METHODS: A social science sub-study of the FEAST (Fluid Expansion As Supportive Therapy) trial. Interviews were held with trial team members (n = 30), health workers (n = 15) and parents (n = 51) from two purposively selected hospitals in Soroti, Uganda, and Kilifi, Kenya. FINDINGS: Overall, deferred consent with prior assent was seen by staff and parents as having the potential to protect the interests of both patients and researchers, and to avoid delays in starting treatment. An important challenge is that the validity of verbal assent is undermined when inadequate initial information is poorly understood. This concern needs to be balanced against the possibility that full prior consent on admission potentially causes harm through introducing delays. Full prior consent also potentially imposes worries on parents that clinicians are uncertain about how to proceed and that clinicians want to absolve themselves of any responsibility for the child's outcome (some parents' interpretation of the need for signed consent). Voluntariness is clearly compromised for both verbal assent and full prior consent in a context of such vulnerability and stress. Further challenges in obtaining verbal assent were: what to do in the absence of the household decision-maker (often the father); and how medical staff handle parents not giving a clear agreement or refusal. CONCLUSION: While the challenges identified are faced in all research in low-income settings, they are magnified for emergency trials by the urgency of decision making and treatment needs. Consent options will need to be tailored to particular studies and settings, and might best be informed by consultation with staff members and community representatives using a deliberative approach.