This paper provides a historical review of the interview research that has been used by science educators to investigate children's basic astronomy knowledge. A wide range of strategies have been developed over the last 120 years or so as successive teams of researchers have endeavoured to overcome the methodological difficulties that have arisen. Hence, it looks critically at the techniques that have been developed to tackle the problems associated with interviews, questionnaires and tests used to research cognitive development and knowledge acquisition. We examine those methodologies which seem to yield surer indications of how young people (at different ages) understand everyday astronomical phenomena--the field often referred to as "children's cosmologies." Theoretical ideas from cognitive psychology, educational instruction and neuroscience are examined in depth and utilised to critique matters such as the importance of subject mastery and pedagogical content knowledge on the part of interviewers; the merits of multi-media techniques; the roles of open-ended vs. structured methods of interviewing; and the need always to recognise the dynamism of memory in interviewees. With illustrations and protocol excerpts drawn from recent studies, the paper points to what researchers might usefully tackle in the years ahead and the pitfalls to be avoided.