We show how reproducibility certification can be used in the context of confidential microdata to increase the quality and the reliability of analyses (both for academic research and statistical studies) based on such restricted-access data. We illustrate this by presenting a partnership between the French restricted data access center (CASD) and cascad, an academic initiative acting as a trusted third-party to conduct reproducibility certification. We present the technical solution which has been deployed and the roadblocks (legal, IT, cultural, etc) that we encountered. Finally we will discuss the cost of conducting reproducible research based on confidential microdata.