Objective: Impaired flow-mediated dilation (FMD) is associated with cardiovascular risk factors and provides prognostic information. Despite the noninvasive nature of this technique, a major limitation to its widespread use is low reproducibility. The aim of this study was to evaluate impact of methodological standardization among different investigation sites on brachial artery FMD reproducibility. Methods: Seven Italian centers recruited 135 healthy volunteers, aged 20-60 years. FMD was assessed by high-resolution ultrasound equipped with a stereotactic probe-holding device. Certified sonographers recorded brachial artery scans at baseline (day 1a), 1 h after (day 1b), and 1 month later (day 30). Endothelium-independent vasodilation (EIVD) to sublingual glyceril-trinitrate was recorded at day 1 and day 30. FMD and EIVD were blindly evaluated at the coordinating center by an automated edge detection system. The intra-session (day la versus 1b) and inter-session (day la versus 30) coefficients of variation were calculated. Results: FMD was not significantly (P = 0.91) different at day 1a, day 1b and day 30 (6.52 +/- 2.9, 6.42 +/- 3.1, 6.57 +/- 2.8%, respectively). The FMD intra-session coefficient of variation was 9.9 +/- 8.4% (from 7.6 to 11.9% across centers). The FMD inter-session coefficient of variation was 12.9 +/- 11.6% (from 11.6 to 16.1% across centers). Inter-session coefficient of variation for EIDV was 19.7 +/- 16.8%. Conclusions: This study shows a homogeneous coefficient of variation for FMD among different centers. The intersession coefficient of variation was similar to the intrasession coefficient of variation, representing the intrinsic FMD variability. We demonstrate for the first time that rigorous and standardized procedure may provide reproducible FMD assessment to study endothelial function in multicenter clinical trials.