Inlet and outlet blood urea concentrations (Cin and Cout) can be used to directly measure dialyzer performance if simultaneous blood flow measurements (Qb) are available. Dialyzer clearance, for example, is the product of the urea extraction ratio [ER = (Cin - Cout)/Cin] and Qb. Urea concentrations are measured routinely in all hemodialysis clinics, but Qb is usually reported as the product of the pump rotational speed and pump segment stroke volume, which can be inaccurate at high flow rates. Dialyzer urea extraction is also a function of Qb, dialysate flow (Qd), and the membrane permeability-area coefficient (K0A) for urea. To determine true in vivo values for Qb and K0A in the absence of direct flow measurements, we developed a model based on an existing mathematical equation for hemodialyzer ER under conditions of countercurrent flow. Qb, K0A, and other variables were adjusted to fit the modeled ER to ER measured in 1,285 patients treated with Qb that ranged from 200 to 450 ml/min during the HEMO Study. Fitting was performed by least squares nonlinear regression using parametric and nonparametric methods for estimating true flow. As Qb rose above 250 ml/min, both methods for estimating actual Qb showed increasing deviations from the flow reported by the blood pump meter. Modeled values for K0A differed significantly among dialyzer models, ranging from 71% to 96% of the in vitro values. The previously described 14% increase in K0A, as Qd increased in vitro from 500 to 800 ml/min, was much less in vivo, averaging only 5.5 +/- 1.5% higher. Dialyzer reprocessing was associated with a 6.3 +/- 1.0% reduction in K0A and an approximate 2% fall in urea clearance per 10 reuses (p < 0.001). Multiple regression analysis showed a small but significant dialysis center effect on ER but no independent effects of other variables, including the ultrafiltration rate, diabetic status, race, ethnicity, sex, method of reuse, treatment time, access recirculation, and use of central venous accesses. The new algorithm allowed a more accurate determination of true Qb and in vivo K0A in the absence of direct flow measurements in a large population treated with a wide range of blood flow rates. Application of this technique for more than 1000 patients in the HEMO Study confirmed that in vitro measurements using simple crystalloid solutions cannot readily substitute for in vivo measurements of dialyzer function, and permitted a more accurate calculation of each patient's prescribed dialysis dose and urea volume.