Malfunction of permanent vascular accesses remains a cause of frequent and costly morbidity in patients receiving chronic hemodialysis (CHD). Several recommendations for routine monitoring of these permanent vascular accesses for incipient failure have been proposed. In this study, multiple indicators of incipient vascular access dysfunction, including "venous" and "arterial" pressures at serial blood flows (200 ml/min, 300 ml/min, and 400 ml/min), percent urea recirculation, Doppler ultrasound, and access blood flow by ultrasound dilution technique were simultaneously evaluated in a total of 220 vascular accesses in 170 chronic hemodialysis patients in two separate study periods (6 months apart). The rate of thrombosis was determined within the subsequent 12 weeks of each study period to assess the short-term predictive power of access thrombosis. During the period of follow-up, there were 34 thrombotic events in 172 polytetrafluoroethylene (PTFE) grafts and only one thrombotic event in 48 arterio-venous fistulas (AVF). Therefore, the statistical analysis was limited to the PTFE grafts. When grafts with thromboses were compared to those without thrombosis by univariate analysis, access blood flow measured either by ultrasound dilution technique (875 +/- 426 ml/min with thrombosis vs. 1193 +/- 677 ml/min without thrombosis, P = 0.001) or by Doppler ultrasound (762 +/- 420 ml/min with thrombosis vs. 1171 +/- 657 ml/min without thrombosis, P = 0.001) were significantly different in the two groups. There was good correlation (r = 0.79, P = 0.0001) between the blood flows determined by both techniques. The grade of stenosis determined by ultrasound was also a statistically significant predictor (P = 0.02). "Venous" and "arterial" pressures were numerically similar and were not statistically different between the accesses that did and those that did not thrombose. When multivariate analysis was used, there was a significantly increased risk of thrombosis only with decreasing access blood flow determined by ultrasound dilution techniques after adjusting for other confounding variables. When the average blood flow of all grafts (1134 ml/min) is considered as the reference access blood flow (relative risk of 1.0), the relative risk of a PTFE thrombotic event within the subsequent 12 weeks was 1.23 at a blood flow 950 ml/min, 1.67 at a blood flow of 650 ml/min and to 2.39 at a blood flow of 300 ml/min. In summary, access blood flow measured by either Dilution or Doppler is a reliable indicator of subsequent short-term thrombosis risk. Other proposed methods of evaluating access dysfunction were not useful in our patients. If simple to use, cost-effective devices to measure dialysis access blood flow become readily available, the measurement of access blood flow will likely become the method of choice for screening of PTFE vascular access dysfunction in hemodialysis patients.