Measurement of the pulmonary-to-systemic blood fl ow ratio (Qp/Qs) in patients with congenital heart disease and shunting lesions may be useful for determining the need for intervention or assessing outcome postprocedure. Frequently used methods to calculate Qp/Qs, such as invasive oximetry, fi rst-pass radionuclide angiography, and Doppler echocardiography have limitations. 1 Phase-velocity cine magnetic resonance imaging (PVC MRI) is a noninvasive technique to measure blood fl ow that has been demonstrated to be accurate in vitro and in vivo. 2– 6 We have previously reported that subjects without a shunt (n 20) had a mean Qp/Qs of 0.99 0.10 by PVC MRI. 3 The aim of this study was to prospectively evaluate the accuracy, reproducibility, and interobserver variability of PVC MRI measurements of Qp/Qs in subjects with a known shunt. Oximetry at catheterization was selected for comparison because this remains the most widely accepted clinical standard. The following criteria were included for entry into this study: (1) diagnosis of either a secundum atrial septal defect or patent foramen ovale, (2) planned cardiac catheterization for device closure of an interatrial communication, (3) anticipated ability to undergo a 30-minute MRI examination without the need for sedation (typically age 7 years). Patients referred for percutaneous device closure of an atrial septal defect or patent foramen ovale were chosen as a subject pool for this investigation because they were already scheduled for cardiac catheterization, and they were more likely to be old enough to undergo MRI without the need for sedation than patients with other shunting lesions. The study was approved by the Committee on Clinical Investigation at Children’s Hospital Boston, and informed, written consent was obtained from all participants or their legal guardian. Subjects underwent MRI to quantify fl ow in the main pulmonary artery (Qp) and ascending aorta (Qs), with 2 separate measurements acquired at each location. In all subjects, all PVC MRI fl ow data were analyzed by 2 operators (AP and BT-G) blinded to each other’s results to determine interobserver variability. MRI studies were performed using a 1.5-Tesla whole-body scanner (Signa LX, General Electric, Milwaukee, Wisconsin) with a torso phased-array radiofrequency receiver coil. Through-plane PVC MRI was performed using a commercially available conventional pulse sequence (Cine PC, General Electric) with retrospective electrocardiographic gating, no respiratory gating, and the following acquisition parameters: echo time 6 to 7 ms, repetition time 18 to 20 ms, fl ip angle 15°, fi eld-of-view 240 to 360 mm, matrix 256 128, slice thickness 6 mm, signal averages 2, and velocity encoding value 150 cm/s for the aorta and 200 cm/s for the pulmonary artery. Imaging time for each measurement was 2.3 to 4.2 minutes depending on heart rate. PVC MRI data were analyzed off-line on a computer workstation (Advantage Windows 4.0, General Electric) equipped with commercially available fl ow analysis software (Flow, MEDIS, Leiden, The Netherlands) including automated vessel border detection. The underlying principles and analysis techniques for PVC MRI have been previously de