Previous studies demonstrated a high incidence of local thrombosis in patients in whom external arteriovenous shunts were used for vascular access. This procedure provides, therefore, a useful model for the evaluation of potential antithrombotic agents. The effect of the hemorheologically and hemostasiologically active drug Pentoxifylline on the incidence of thrombosis of arteriovenous shunts (Ramires shunt) was investigated in a long-term, double-blind, placebo-controlled study in 51 patients on chronic hemodialysis. The two treatment groups were comparable in age, sex, concomitant medication, and dialysis program (three times per week for four hours). Drugs known to affect platelet function or coagulation were excluded, with the exception of heparin, during the dialysis procedure. All shunts were placed in the forearm and inserted into the distal part of the radial artery and basilic antebrachial vein. Simultaneously, for medical reasons, in all patients an arteriovenous fistula was performed (proximal part of radial artery and cephalic antebrachial vein). Shunt thrombosis was assumed when the flow in the shunt discontinued under visual and auscultatory control. Thrombi were documented by physical removal from the arterial part of the shunt by use of gentle suction or by complete shunt thrombosis (both arterial and venous part of the shunt). Thereafter, the patients' trial period terminated. The total number of thrombi during the observation period was 44 in the pentoxifylline group (26 patients), compared with 82 in the placebo group (25 patients). The mean number of thrombi per patient was 1.69 +/- 1.29 in the pentoxifylline group, significantly lower than that in the placebo group (3.28 +/- 1.99/p less than 0.05).(ABSTRACT TRUNCATED AT 250 WORDS)