The possibility of increasing the resistance of a synthetic vascular graft to intraoperative or immediate postoperative bacterial contamination justifies the interest in methods such as antibiotic bonding or antibiotic soaking. On the basis of this experience and with the aim of testing the efficacy of such a graft, a multicentre experimental study on sheep to compare the susceptibility to infection of Dacron Gelseal grafts (control) versus rifampicin-bonded Dacron Gelseal grafts (treated) following the intravenous infusion of 10(7)-10(8) cells Staphylococcus aureus was conducted. The grafts were implanted in both common carotid arteries of sheep. In a group of 11 animals (group A), a treated and a control graft were implanted in the same animal. In a group of four sheep (group B), only treated or control grafts were implanted in each animal. In group A, 36% of (four of 11) the treated grafts became infected versus 54% (six of 11) of the control prostheses. In group B, none of the treated grafts was infected by the inoculated pathogen, which, by contrast, infected 75% (three of four) of the controls. These observations confirm the recent interest aroused by the possibility of pretreating gelatin-coated Dacron grafts with rifampicin in the prevention of early graft infection.