The expansion of procedures in multidisciplinary vascular medicine has sparked a controversy regarding measures of quality improvement. In addition to primary registries, the use of health insurance claims data is becoming of increasing importance. However, due to the fact that health insurance claims data are not collected for scientific evaluation but rather for reimbursement purposes, meticulous validation is necessary before and during usage in research and quality improvement matters. This review highlights the advantages and disadvantages of such data sources. A recent comprehensive expert opinion panel examined the use of health insurance claims data and other administrative data sources in medicine. Results from several studies concerning the validity of administrative data varied significantly. Validity of these data sources depends on the clinical relevance of the diagnoses considered. The rate of implausible information was 0.04 %, while the validity of the considered diagnoses varied between 80 and 97 % across multiple validation studies. A matching study between health insurance claims data of the third-largest German health insurance provider, DAK-Gesundheit, and a prospective primary registry of the German Society for Vascular Surgery demonstrated a good level of validity regarding the mortality of endovascular and open surgical treatment of abdominal aortic aneurysm in German hospitals. In addition, a large-scale international comparison of administrative data for the same disorder presented important results in treatment reality, which differed from those from earlier randomized controlled trials. The importance of administrative data for research and quality improvement will continue to increase in the future. When discussing the internal and external validity of this data source, one has to distinguish not only between its intended usage (research vs. quality improvement), but also between the included diseases and/or treatment procedures. Linkage between primary registry data and administrative data could be a reasonable solution to some current major issues of validity.