OBJECTIVE The variability of consecutive cine phase-contrast MR flow measurements could significantly affect their use for clinical decisions, especially during provocative testing. The purposes of this study were to determine the normal variability of flow and consecutive flow measurements in the main portal vein on MR images and to determine how intraobserver variability, interobserver variability, and MR imager variability affect these measurements. METHODS MR flow measurements were acquired four consecutive times at the same location in the main portal vein of 12 subjects and three consecutive times at the same location in a nonpulsatile vessel model. All acquisitions were completed within 10 min. All main portal vein MR data sets were evaluated manually in a blinded review by two independent observers during three separate sessions spaced a mean of 4.5 weeks apart. Flow model data sets were evaluated during a single session by one observer. Variabilities were subsequently calculated by a components-of-variance analysis and by the coefficient of variation (SD/mean x 100). RESULTS Of the total variance, 90% was due to flow variability among subjects (intersubject), 6% to flow variability within one subject (intrasubject), 2% to intraobserver variability, and 2% to interobserver variability. The coefficient of variation of consecutive MR portal vein flow measurements within a single subject was 11% +/- 5% (range, 3-23%). Intra- and interobserver variabilities were 5% +/- 2% (range, 1-11%) and 4% +/- 4% (range, 0-17%), respectively. MR imager variability was 1% +/- 1% (range, 0-2%). CONCLUSIONS The mean variability of consecutive cine phase-contrast MR flow measurements in the main portal vein is 11% +/- 5% and could affect research and clinical protocols that employ this technique.