Introduction: Holodiastolic flow reversal (HDR) in the descending aorta (DA) may indicate hemodynamically significant aortic regurgitation (AR). However, the optimal definition and quantification of flow reversal on cardiac magnetic resonance (CMR) is unknown. Methods: We studied 346 chronic AR patients with regurgitant volume (RegVol) ≥30 mL or regurgitant fraction (RegF) ≥30%. The presence of HDR and quantitation of reverse volume in the DA (DA-RVol) was performed by a blinded observer. The threshold for DA-RVol that identified ≥ moderate-severe AR (RegVol ≥45 mL and RegF ≥40%) was established by derivation (50% random sample), validation (50% remaining sample), and ROC curve analysis. Multivariable logistic regression models adjusting for age, gender, NYHA Class, and indexed left ventricular end-diastolic volume were performed to compare HDR and DA-RVol. Results: The median age was 61 (IQR 50-70) years, 79.2% men, and 30.3% with bicuspid aortic valve. The median RegF was 37% (IQR 32-44) and DA-RVol was 12 mL (IQR 5-27). HDR was observed in 150 (43.2%), with 63% sensitivity and 77% specificity for detecting ≥ moderate-severe AR. The threshold of DA-RVol which identified ≥ moderate-severe AR was 17 mL (AUC 0.82, sensitivity 68%, and specificity 87%). The ROC curve analysis revealed a higher AUC for DA-RVol ≥17 mL relative to HDR (0.78 vs 0.70, P <0.001) for predicting ≥ moderate-severe AR. On multivariable analysis models, DA-RVol ≥17 mL was superior to HDR (AUC 0.83 vs 0.80, P =0.008). Conclusion: Quantitation of flow reversal in the descending aorta is a more reliable method to identify hemodynamically significant AR compared to simple presence of holodiastolic flow reversal.