In this paper a combination of methods is used to investigate the determinants of birth-interval length in a single city in Burkina Faso. First, a multi-round survey provided accurate measurement of dates of reproductive events. Secondly, open-ended interviews with a sub-sample of women from the survey were tape-recorded and transcribed. In order to disentangle the direct and indirect biological and behavioural determinants of birth-interval length, seven series of hazards models were estimated. These results were compared with women's perceptions of the same relationships revealed in the in-depth interviews. The lesson we draw is that no single data-collection strategy can provide answers to all questions of interest. The multi-round survey provides accurate information on actual behaviour, but at great cost. The in-depth interviews enable the investigator to study norms and motivations, but this method is impractical for data collection for a large representative sample. Combining the two strategies can yield insights that may enhance the effectiveness of programme interventions.