Flow assessment with phase contrast magnetic resonance imaging (PC-MRI) protocols is an important component of a comprehensive cardiovascular MR (CMR) assessment. Breath-hold (BH) and non-breath-hold (NBH) PC-MRI protocols are widely available for this imaging modality. Because flow in the great vessels is known to vary with the respiratory cycle, we hypothesized that these 2 approaches might yield different results in the clinical assessment of forward and regurgitant flow in the ascending aorta. Further, given renewed awareness of the possible effect of velocity offsets in PC-MRI, we also sought to evaluate the impact of BH and NBH protocols on this potential source of error. A prospective observational study was performed in 55 consecutive patients referred for clinical CMR of the thoracic aorta. Both BH and NBH protocols were performed at the sinotubular junction and at the mid ascending aorta. Ten additional patients underwent repeated scanning at the mid ascending aorta with both BH and NBH protocols so that protocol variability could be assessed. Finally, ten patients were scanned with both BH and NBH protocols, and phantoms were then imaged with identical imaging parameters so that offset errors associated with each protocol could be evaluated. Forward flow was generally greater with the NBH protocol than with the BH protocol (mean values 102.1 mL vs. 97.9 mL; P = 0.0004). The Bland-Altman limits of agreement were quite wide for all indices (e.g, forward flow, -26.7 mL, +18.2 mL), which suggests that results from BH and NBH protocols cannot be interchanged with confidence. Estimated phase offset errors were similar for both protocols and were generally within acceptable ranges at the mid ascending level, with slightly higher values observed at the sinotubular junction for the BH technique. We observed differences in flow values with BH and NBH protocols for PC-MRI. This finding is relevant to patients imaged serially for the evaluation of cardiac output or valve (aortic or mitral) insufficiency, for whom adherence to one PC-MRI breathing protocol is likely most effective.