The consistency of the determination of A-V fistula recirculation (R) using the thermodilution method (T) with a new probe (blood temperature monitor, BTM Fresenius A.G.) was studied in 32 patients (AVF: proximal 34%, distal 63%, graft 3%). We compared R calculated by T with both the traditional three-sample method (C) and the low-flow three-sample method (L); both BUN and creatinine (CR) were measured in all samples at the beginning and at the end of the session. T was also determined at the 2nd and 3rd hour. There was a significant correlation between T and either C or L at the start of the session (BUN and CR) as well as at the end (only CR). R was higher (11.9 +/- 10) in proximal AVF than in the distal (5 +/- 3.1%; p0.01) when measured by T at the same blood flow (QB: 313 +/- 45 vs 343 +/- 52 mls/min, p = ns). T increased but not significantly by increasing Qb from 150 to 300 mls/min in ten patients. No correlation was found during the session between blood pressure and T variations. In conclusion, T and L give very similar results while C overestimates recirculation. R is easy to perform repeteadly by T with results available online.