Measuring blood flow in arteriovenous fistulae in patients under chronic dialysis is of interest to evaluate the repercussions of the fistula on the heart. The apparently simplest method is direct measurement of the mean blood flow velocity by the pulsed doppler technique and ot the cross-section area by ultrasonography, the product of these two values being the blood flow rate. Another method has been proposed, which consists of measuring the cardiac output before and after compression of the fistula, the difference between the two values being supposed to represent the blood flow rate in the fistula. A comparative study of these two methods was conducted in 17 patients aged from 2 to 21 years (mean: 14 years). The direct method gave a figure of 475 ml.min-1.m-2 (SD = 240), while the figure obtained with the indirect method was 471 ml.min-1.m-2 (SD = 227); the difference was statistically not significant. In terms of concept, however, the indirect method is open to much more severe criticism than the direct method, and whenever possible the latter should be preferred when measuring blood flow in arteriovenous fistulae.