Three-dimensional velocity-encoded cine magnetic resonance imaging (3D VEC MRI) allows for calculation of secondary flow parameters that may be used to estimate prognosis of individual cardiovascular diseases. However, its accuracy has not been fully investigated yet. The purpose of this study was to validate aortic flow quantification by 3D VEC MRI in vitro and in vivo using stacked two-dimensional acquisitions. Time-resolved stacks of two-dimensional planes with three-directional velocity-encoding (stacked-2D-3dir-MRI) were acquired in an elastic tube phantom with pulsatile flow simulating aortic flow as well as in 11 healthy volunteers (23 ± 2 years). Previously validated two-dimensional through-plane VEC MRI at six equidistant levels in vitro and three locations in vivo (ascending aorta/aortic arch/descending aorta) was used as reference standard. The percentage difference of the stacked-2D-3dir-MRI measurement to the reference standard was defined as the parameter for accuracy. For in vitro aortic flow, stacked-2D-3dir-MRI underestimated average velocity by −6.8% (p < 0.001), overestimated average area by 13.6% (p < 0.001), and underestimated average flow by −7.4% (p < 0.001). Accuracy was significantly higher in the field of view centre compared to off-centre (p = 0.001). In vivo, stacked-2D-3dir-MRI underestimated average velocity (all three locations p < 0.001) and overestimated average area at all three locations (p = n.s./<0.001/<0.001). Average flow was significantly underestimated in the ascending aorta (p = 0.035), but tended to be overestimated in the aortic arch and descending aorta. In conclusion, stacked-2D-3dir-MRI tends to overestimate average aortic area and to underestimate average aortic velocity, resulting in significant underestimation of average flow in the ascending aorta.