Background: Missing and inconsistent data are common problems when researching young people’s sexual health. Complete and high-quality data are needed to inform effective programming. To inform improvements, we examine different techniques for collecting school-based data about young people’s sexuality, relationships and partner violence.Methods: We use empirical data from a school-based programme evaluation in Mexico City as a case study. The evaluation sought data on sexuality and relationships using self-administered questionnaires, observation, focus groups, and in-depth interviews. We explore the advantages and disadvantages of different data collection methods in practice, comparing data from questionnaires and interviews to examine data quality and identifying challenges to collecting and interpreting data from different methods and sources. We used descriptive statistics, reviewed field notes, and conducted thematic analysis, drawing on research team and participant perspectives. Results: Data collection was influenced by earthquakes and extreme weather, social aspects of the data collection setting, and the sensitive nature of the data. There was variation in data quality by group, timepoint, gender, and topic. Women had a higher proportion of complete responses than men at most timepoints. Intervention group men were the only group to increase their proportion of complete responses from baseline to endline. Items about sexual activity had the lowest proportions of complete responses. Participants varied in their willingness to share personal information. Data from different sources and timepoints sometimes appeared contradictory, creating challenges with interpretation. Conclusions: This analysis can inform methodological choice for school-based sexual health research. Our experience highlights the difficulties of anticipating – and correcting for – the many interacting real-world challenges threats to data quality and interpretation challenges that can arise. It is vital that results for all types of analysis are accompanied by a reflexive discussion of data collection conditions and challenges that might impact on data quality.