Complex and lengthy interventional radiological techniques have resulted in a number of patients developing skin reactions in recent years. To safeguard against these side effects, we have investigated the degree to which entrance skin dose can be reduced by inserting 0.18 mm and 0.35 mm copper filtration in the incident beam. The potential reduction was measured on a 22 cm water phantom for each of eight models of a fluoroscopy unit. Using the catheter laboratory fluoroscopy unit on which radiofrequency ablations are routinely performed, we assessed the relative effectiveness of adding filtration and increasing the kV:mA ratio. Image quality was subjectively assessed for diagnostic and therapeutic acceptability in two groups of 10 patients undergoing radiofrequency ablations, pacemaker insertions or electrophysiology studies. One of the groups was screened with 0.35 mm copper filtration in place and the other group acted as the control. Maximum patient skin dose proved difficult to measure directly because of the unpredictable dose pattern. This pattern was studied in four patients using a film method in conjunction with thermoluminescent dosemeters. Copper filtration 0.35 mm thick inserted in the beams of the eight fluoroscopy units produced a mean reduction in entrance dose to the phantom of 58% with a mean increase in tube loading of 29%. At 100 kV the increased loading on the X-ray tube was equivalent to increasing the anteroposterior separation of the patient by 2 cm. Measurements on the catheter laboratory unit showed that the tube voltage would need to be raised above the normal diagnostic range to obtain an equivalent entrance dose reduction without the filter. The blackening of films under the patients showed complex patterns, but the estimated skin doses were consistent with those predicted by the phantom experiments. All six cardiologists considered there to be insignificant detriment to image quality in the procedures investigated.