OBJECTIVE: To assess the outcome of arterial duct occlusion with coils chosen according to the duct morphology. DESIGN: Retrospective study. SETTING: Paediatric cardiology centre. PATIENTS: Coil occlusion was attempted in 57 patients aged 0.5 to 15 (median 3.7) years and weighing 5-59 (median 14) kg between January 1991 and December 1995. A residual leak was present in 8 patients after umbrella closure and in 4 patients after duct ligation. METHODS: Coils of 4 different types were implanted through 4 or 5 F femoral artery catheters. Platinum or Interlocking Detachable 0.018 inch coils were deployed completely inside tubular ducts. Gianturco or PDA controlled release 0.038 inch coils were implanted to straddle short, post ligation and post umbrella ducts. RESULTS: Coil implantation was successful in 54/57 patients. At 1 year the cumulative occlusion rate was 53/57 ducts (93%) on an intention to treat analysis. A single coil was implanted in 37 (69%), 2 coils in 10 (19%), 3 coils in 3 (5%) and 4 coils in 4 (7%) of the 54 successful procedures. Duct occlusion was documented at the end of the procedure in 31%, by the following day in 83%, by 6 weeks in 87%, by 6 months in 96%, and by 1 year in 98%. Coil embolisation occurred in 6/58 procedures (10%), with a 50% rate in the first year of implantation (1/2 patients) falling to 7% in the last year (3/42 patients). All embolised coils were easily retrieved. CONCLUSIONS: Occlusion of small to moderate size arterial ducts, including residual post umbrella or post ligation ducts, was readily accomplished by coils selected according to the duct anatomy. This has both cost and practical benefits.