This paper describes an evaluation study that aimed to assess data collection processes in a population-based case–control study of very preterm birth. Semi-structured interviews were conducted with the 10 research interviewers to determine their perceptions of the hypotheses, the differences between interviewing cases and controls and between modes of interview, their reactions to questions which they had to ask in interviews and their training. Time and cost of the collection of data were also considered. None of the research interviewers identified which of the questions in the interview constituted the primary hypothesis. All interviewers interviewed cases and controls (including mothers of twins and singletons), and collected data face-to-face and by telephone. Whilst half of the interviewers had no issue with asking sensitive questions, hearing of intimate partner violence, risk-taking behaviour and inappropriate medical care were confronting for others. Training was judged as adequate, as was the continuing support of the project co-ordinator. On average, interviewing took 40% of research interviewers’ time, and 25 interviews were completed per effective full-time month of interviewer time. Each interview cost approximately $170AU to complete. In relation to this case–control study, interviewer bias may be lessened since research interviewers had been unable to infer the hypothesis of the study. All interviewers had interviewed both cases and controls and any systematic differences can be adjusted for in the analysis.