In assessing the value of different approaches to real-time monitoring for the most vulnerable, an initial requirement is to set out a conceptual framework that provides at least some degree of clarity as to what precisely is meant by ‘real time’, ‘monitoring’ and ‘vulnerable’– all terms that can be highly context-specific. That is the first task addressed here. The second is to consider potential sources of data that might be used to undertake real-time monitoring and assess their advantages and disadvantages for the present purpose. Four general approaches are considered – community-based participatory monitoring, sentinel sites, routine data systems and rapid surveys – and selected examples from the literature are given to illustrate the potential use and limitations of their applications.1