Agile software development, despite its novelty, is an important domain of research within software engineering discipline. Agile proponents have put forward a great deal of anecdotal evidence to support the application of agile methods in various application domains and industry sectors. Scientifically grounded empirical evidence is, however, still very limited. Most scientific research to date has been conducted on focused practices performed in university settings. In order to generate impact on both the scientific and practical software engineering community, new approaches are needed for performing empirically validated agile software development studies. To meet these needs, this paper presents a controlled case study approach, which has been applied in a study of extreme programming methodology performed in close-to-industry settings. The approach considers the generation of both quantitative and qualitative data. Quantitative data is grounded on three data points (time, size, and defect) and qualitative data on developers’ research diaries and post-mortem sessions.