Background: Cardiovascular magnetic resonance (CMR) is considered the gold standard of cardiac volumetric measurements. Flow in the aortic root is often measured at the sinotubular junction, even though placing the slice just above valve level may be more precise. It is unknown how much flow measurements vary at different levels in the aortic root and which level corresponds best to left ventricle volumetry. Methods: All patients were older than 70 years presenting with at least one of the following diagnoses: diabetes, hypertension, prior stroke and/or heart failure. Patients with arrhythmias during CMR and aortic stenosis were excluded from the analyses. Stroke volumes were measured volumetrically (SVref) from steady-state free precision short axis images covering the entire left ventricle, excluding the papillary muscles and including the left ventricular outflow tract. Flow sequences (through-plane phase contrast velocity mapping) were obtained at valve level (SVV) and at the sinotubular junction (SVST). Firstly, SVV and SVST were compared to each other and secondly, after excluding patients with mitral regurgitations to ensure that stroke volumes measured volumetrically would theoretically be equal to flow measurements, SVV and SVST were compared to SVref. Results: Initially, 152 patients were included. 22 were excluded because of arrhythmias during scans and 9 were excluded for aortic stenosis. Accordingly, data from 121 patients were analysed and of these 63 had visually evident mitral regurgitation on cine images. On average, stroke volumes measured with flow at the sinotubular junction was 13-16 % lower than when measured at valve level (70.0 mL ±13.8 vs. 81.8 mL ±15.5). This was in excess of the expected difference caused by the outflow to the coronary arteries. In the 58 patients with no valvulopathy, stroke volumes measured at valve level (79.0 mL ±12.4) was closest to the volumetric measurement (85.4 mL ±12.0) but still significantly lower (p < 0.001). Flow measured at the ST-junction (68.1 mL ±11.6) was significantly lower than at valve level and the volumetric measurements. The mean difference between SVref-SVV (6.4 mL) and SVref-SVST (18.2 mL) showed similar variances (SD 7.4 vs. 8.1 respectively) and hence equal accuracy. Conclusions: Aortic flow measured at valve level corresponded best with volumetric measurements and on average flow measured at the sinotubular junction underestimated flow approximately 15 % compared to valve level. Trial registration: ClinicalTrials.gov identifier: NCT02036450. Registered 08/01/2014.