Résumé : Objective: To report short- and long-term effects of an audit process intended to optimise the radiation dose from multi-detector row computed tomography (MDCT). Methods: A survey of radiation dose from all eight MDCT departments in the state of Luxembourg performed in 2007 served as baseline, and involved the most frequently imaged regions (head, sinus, cervical spine, thorax, abdomen, and lumbar spine). CT dose index volume (CTDIvol), dose-length product per acquisition (DLP/acq), and DLP per examination (DLP/exa) were recorded, and their mean, median, 25th and 75th percentiles compared. In 2008, an audit conducted in each department helped to optimise doses. In 2009 and 2010, two further surveys evaluated the audit's impact on the dose delivered. Results: Between 2007 and 2009, DLP/exa significantly decreased by 32-69% for all regions (P <0.001) except the lumbar spine (5%, P =0.455). Between 2009 and 2010, DLP/exa significantly decreased by 13-18% for sinus, cervical and lumbar spine (P ranging from 0.016 to less than 0.001). Between 2007 and 2010, DLP/exa significantly decreased for all regions (18-75%, P <0.001). Collective dose decreased by 30% and the 75th percentile (diagnostic reference level, DRL) by 20-78%. Conclusions: The audit process resulted in long-lasting dose reduction, with DRLs reduced by 20 -78%, mean DLP/examination by 18-75%, and collective dose by 30%. Key points: • External support through clinical audit may optimise default parameters of routine CT. • Reduction of 75th percentiles used as reference diagnostic levels is 18 -75%. • The effect of this audit is sustainable over time. • Dose savings through optimisation can be added to those achievable through CT. © European Society of Radiology 2013.