Following the release of European Directive EU 97/43, radiodiagnostic facilities within the European Union are required to implement a system of patient dose reviews based on comparisons with European, national and local diagnostic reference levels (DRLs). Establishing these levels for typical interventional radiology examinations presents a problem as definition of 'typical' examinations can be difficult, patient numbers are limited and these procedures are often performed at a few specialist centres. This paper uses dose-area product (DAP) gathered over a period of 3 years from 40 fluoroscopy rooms to investigate potential difficulties when it comes to forming diagnostic reference levels for interventional radiology. Comparison of DAP distributions with standard complex (fluoroscopy based) examinations such as barium enema reveals considerably more variation for interventional procedures. Two methods of forming a DRL are compared: pooled patient DAP distributions versus a distribution of DAP per room. The bootstrap resampling method is then applied to DAP distributions to form a confidence interval for the chosen DRL statistic. Potential error on a DRL formed at a local level from a limited number of patient dose readings and x-ray rooms is significant. The results are reviewed in the wider context of DRLs in general radiology. For complex examinations, it is suggested that the function of the DRL is best served by setting DRLs based on pooled size-corrected patient DAP distributions rather than distributions of average DAP per room.