Satisfactory hemodialysis access flow (Qa) is necessary for dialysis adequacy. However, high access flows are postulated to increase cardiac output (CO). The relationship between Qa and CO is not well defined. The purpose of this study was to observe the relationship between Qa and CO and to evaluate the effect of blood volume change (BVdelta) on Qa and CO during hemodialysis (HD). Measurements of Qa and CO (ultrasound dilution; Transonics Monitor, Ithaca, NY) were performed sequentially at baseline in 18 patients (13 forearm arteriovenous fistulae, 5 Gore-Tex grafts) and after an intervention involving either HD with attempted zero BVdelta (mean: -0.4%; range: -2.6 to 1.6%) or a significant BVdelta (mean: -7.3%; range: -3.1 to -11.9%). Measurement of BVdelta was done by hematocrit dilution (Crit-Line Monitor, In-Line Diagnostics, Riverdale, UT). The volume ultrafiltered (V(UF)) and the mean arterial pressure (MAP) were recorded at baseline and after intervention. In five patients with fistulae, CO was measured after manual occlusion of the fistula for 1 min. At baseline, mean (+/-SD) Qa was 1455+/-600 ml/min, and CO was 6.8+/-1.8 L/min. The relationship between Qa and CO was strong, Qa = 0.20 CO + 0.06 (r = 0.62; p = 0.01); this was not significantly altered with either intervention. Access flow was not changed with either zero BVdelta or significant BVdelta. Cardiac output was not altered when there was no BVdelta; however, CO did decrease by 1.2+/-0.6 L/min (p<0.001) after BVdelta reduction. The Qa/CO ratio was unchanged after zero BVdelta but was increased after BVdelta (p = 0.004). There were no correlations with MAP change or V(UF). There were no differences in Qa, CO, or Qa/CO by access type. The mean Qa/CO was 21+/-6%. Three patients had Qa/CO <15%, and they all had access stenoses. Cardiac output did not decrease after transient (1 min) occlusion of the fistula. In conclusion, there is a strong relationship between Qa and CO. With BVdelta, the Qa is maintained while the CO falls and the Qa/CO increases, perhaps by reflex vasoconstriction of the systemic circulation. Longitudinal studies are required to determine which is the dependent variable. A low Qa/CO may indicate access dysfunction.