Color flow doppler ultrasound examination of the hemodialysis access was conducted in 2,792 hemodialysis patients to evaluate its value in predicting hemodialysis access failure. After baseline assessment of vascular access function with clinical and laboratory tests including color flow doppler evaluation these patients were followed for a minimal of 6 months or until graft failure occurred (defined as surgery or angioplasty intervention, or graft loss). The patient demographics and vascular accesses were typical of a standard hemodialysis patient population. On the day of the color flow doppler examination systolic and diastolic blood pressure, hematocrit, urea reduction ratio, dialysis blood flow, venous line pressure at a dialysis blood flow of 250 ml/min, and access recirculation rate were measured. At the conclusion of the study 23.5% of the patients had access failure. Case mix predictors for access failure were determined using the Cox Model. Case mix predictors of access failure were race, non-white was higher than white (p < 0.005), younger accesses had a higher risk than older accesses (p < 0.025), accesses with prior thrombosis had a higher risk of failure (p = 0.042), polytetrafluoroethylene (PTFE) grafts had a higher risk than native vein fistulae (p < 0.05), loop PTFE grafts had a higher risk than straight PTFE grafts (p < 0.025), and upper arm accesses had a higher risk than forearm accesses (p = 0.033). Most significant, however, was decreased access blood flow as measured by color flow doppler (p < 0.0001). The relative risk of graft failure increased 40% when the blood flow in the graft decreased to less than 500 ml/min and the relative risk doubled when the blood flow was less than 300 ml/min. This study has shown that color flow doppler evaluation, quantifying blood flow in a prosthetic graft, can identify those grafts at risk for failure. In contrast, color doppler volume flow in native AV fistulae could not predict fistula survival. This technique is noninvasive, painless, portable, and reproducible. We believe that preemptory repair of an anatomical abnormality in vascular access grafts with decreased blood flow may decrease patient inconvenience, associated morbidity, and associated costs.