Blood recirculation is one of the key factors of decreasing dialysis efficiency. Determination of recirculation rate (R) is necessary to optimize effective dialysis delivery and to monitor vascular access function. R can be directly measured by a conductivity method in paired filtration dialysis (PFD), a double-compartment hemodiafiltration system that permits direct access to plasma water via the ultrafiltration stream. Measurement of R, in this system, involves the first of two conductivity sensors integrated in a urea monitor (UMS, Bellco-Sorin, Mirandola, Italy), and two saline injections. The rise in conductivity (δC1) induced by a 2.7 ml bolus of hypertonic saline 20% (mg/dl) in the arterial line serves for calibration, and is followed by an equivalent injection into the venous line, giving rise to δC2. The ratio δC2/δC1 equals R. A comparison between R values obtained with this method and with the low-flow technique in 31 chronic dialysis patients during 138 PFD sessions is reported. Mean R ± SD by the conductivity method was 5.1 ± 2.0 and 5.7 ± 2.0% after 65 and 155 minutes of PFD (correlation coefficient, r = 0.75), whereas it was 6.4 ± 4.9% and 5.5 ± 4.6% after 30 sec of low blood pump flow for urea and creatinine markers, respectively (r = 0.35). After 120 sec of low flow, mean R increased to 9.0 ± 5.1 and 8.8 ± 4.6% for urea and creatinine, respectively (r = 0.45). Considerable discrepancies were found in R values measured simultaneously with the two blood markers. Statistically significant differences were found between the two measurement modalities (blood-side and conductivity); the correlation coefficients (r) varied between 0.28 and 0.41. The observed differences in mean R results do not seem considerable from a clinical perspective. The best agreement between blood-side and conductivity R measurements was obtained with Rcreat after 30 sec of low flow. Overall, a wider distribution was found in R values from blood-side determinations, most likely consequent to variability in the dosing method. The conductivity method appears more accurate and simple in assessing total R, and can be readily automated and integrated into the dialysis machine. The authors, therefore, recommend evaluation of R using methods not based on chemical blood concentration values.