Thromboembolic complications have been attributed to the use of radiographic contrast media (CM) during interventional procedures for arterial revascularization. However, due to the low frequency of adverse events, comparisons between different CM have been difficult to perform, although it has been suggested that ionic (vs. non-ionic) CM may be associated with fewer thrombotic events. The present study was undertaken using well-characterized baboon thrombosis models in order to compare different CM under physiologically relevant and controlled conditions of blood flow, exposure time, and CM concentration. Three CM were studied: ioxaglate, iohexol, and iodixanol. CM were locally infused into the proximal segment of femoral arteriovenous shunts. Palmaz-Schatz stents (4 mm i.d.) and expanded tubular segments (9 mm i.d.), which exhibited venous-type flow recirculation and stasis, were deployed into the shunts distally. Saline was infused in identical control studies. Blood flow was maintained at 100 ml/min. Thrombosis was measured over a blood exposure period of 2 hours by gamma camera imaging of 111In-platelets and by gamma counting of deposited 125I-fibrin. CM concentrations within the flowfield were predicted using computational fluid dynamics. At infusion rates of 0.1 and 0.3 ml/min, the low-osmolar ionic CM ioxaglate reduced both platelet and fibrin deposition on the stents by 75-80% (p <0.005), while both iohexol and iodixanol reduced platelet deposition by 30-50% (p <0.05). In the regions of low shear flow, ioxaglate (0.3 ml/min) also reduced platelet deposition significantly (by 52% vs. control results; p <0.05). Thus the three agents evaluated--ioxaglate, iohexol, and iodixanol--all produced anticoagulant and antiplatelet effects and were inherently antithrombotic in vivo. The most striking effects were seen with the low osmolarity, ionic contrast agent ioxaglate.