The brachial artery flow-mediated dilation test (FMD) is the non-invasive gold-standard used to assess endothelial function. Reduced FMD is an early event in the development of atherosclerosis and provides a marker for predicting future cardiovascular disease events. Despite its widespread popularity and proven validity, the FMD test is limited by poor reliability. There are three major limitations associated with the standard FMD methodology: (i) inappropriate expression of FMD, (ii) measurement variance associated with a short-lived FMD response and (iii) most studies fail to account for the FMD stimulus. A series of relatively simple statistical practices can be adopted to account for these limitations. In particular, we suggest that endothelial function be estimated using shear rate-diameter dose-response curves, which can be statistically analysed using hierarchical linear modelling. The use of dose-response curves could potentially improve measurement reliability and validity.