BACKGROUND The aim of this study was to determine whether the routine use of an antibiotic-bonded gelatin-coated Dacron graft could reduce the incidence of prosthetic graft infection. Extra-anatomic grafts were chosen for study as they have the highest risk of graft infection. This paper reports early results up to 1 month after surgery. METHODS This multicentre study involved 14 vascular units in the UK. A total of 257 patients underwent extra-anatomic bypass. Patients were randomized to rifampicin bonding (1 mg/ml rifampicin soak for 15 min before graft insertion) or a control group. Routine three-dose antibiotic prophylaxis was administered to patients in both groups. RESULTS There were 178 men and 79 women of median age 69 (range 43-92) years. Rifampicin-bonded (n=123) and control (n=134) groups were well matched for clinical details, risk factors and operative techniques. No side-effects were noted from rifampicin bonding. Only one patient (in the control group) developed a graft infection and this proved fatal. There were no significant differences between bonded and unbonded grafts in terms of perioperative mortality rate (9 and 5 per cent respectively), median hospital stay (10 days for both groups), total infective complications (15 and 21 per cent respectively) or need for postoperative antibiotics (13 and 18 per cent respectively). CONCLUSIONS Early results from this study have not identified any significant advantage in the routine use of rifampicin bonding, but the rate of graft infection was very low (0.4 per cent). Gelatin coating alone may provide protection against infection. Definitive recommendations about the role of antibiotic bonding cannot be made until longer follow-up becomes available.