Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.Objectives: Brachial artery flow-mediated dilation (FMD) is a popular technique to examine endothelial function in humans. Identifying volunteer and methodological factors related to variation in FMD is important to improve measurement accuracy and applicability. Methods: Volunteer-related and methodology-related parameters were collected in 672 volunteers from eight affiliated centres worldwide who underwent repeated measures of FMD. All centres adopted contemporary expert-consensus guidelines for FMD assessment. After calculating the coefficient of variation (%) of the FMD for each individual, we constructed quartiles (n = 168 per quartile). Based on two regression models (volunteer-related factors and methodology-related factors), statistically significant components of these two models were added to a final regression model (calculated as ß-coefficient and R 2). This allowed us to identify factors that independently contributed to the variation in FMD%. Results: Median coefficient of variation was 17.5%, with healthy volunteers demonstrating a coefficient of variation 9.3%. Regression models revealed age (ß = 0.248, P <0.001), hypertension (ß = 0.104, P <0.001), dyslipidemia (ß = 0.331, P <0.001), time between measurements (ß = 0.318, P <0.001), lab experience (ß = -0.133, P <0.001) and baseline FMD% (ß = 0.082, P <0.05) as contributors to the coefficient of variation. After including all significant factors in the final model, we found that time between measurements, hypertension, baseline FMD% and lab experience with FMD independently predicted brachial artery variability (total R2 = 0.202). Conclusion: Although FMD% showed good reproducibility, larger variation was observed in conditions with longer time between measurements, hypertension, less experience and lower baseline FMD%. Accounting for these factors may improve FMD% variability.