Aims: Accurate evaluation of aortic size and growth is crucial in aor-tic disease. We aimed to compare different MRI acquisition strategies without contrast agent injection, and define the optimal balance between scan time and reliable quantification of 3D aortic morphology. Methods: 48 healthy volunteers (39±14years, 29women) underwent non-contrast respiratory- and ECG-gated MRI (1.5T, Siemens) of the aor-ta in sagittal view during diastole. A gradient-echo sequence with spec-tral fat saturation (fs), spatial resolution (SR)=1.28x1.42x2mm3, and ac-quisition time=7:42±2:27min was used as reference. Then, 3 strategies based on compressed sensing (CS, research prototype) with isotropic SR=1.3mm3 were tested while using either fs (CSfs, acceleration factor R=5), inversion-recovery (CSSPAIR, R=5) or Dixon (CSDixon, R=7). A last acquisition was performed using CS, R=5, fs and SR=1.28x1.42x2mm3 (CSresoREF). Aortic cross-sectional diameter at 7 anatomical landmarks from sino-tubular junction to celiac trunk, length and volume of ESC ref-erence segments #1-4 were measured from each dataset, after semi-automated 3D aortic segmentation using custom software (Mimosa), blinded to acquisition strategy. Results: As expected, CS strategies were significantly faster than fs reference (CSfs: 2:56±0:51min; CSSPAIR: 3:03±0:59min, CSDixon: 3:12±0:55min, CSresoREF: 1:38±0:32min; all p<0.0001). Image quality and aortic segmentation were visually higher for reference (0 exclusion) than for CS acquisitions (diameter/length and volume measurements were ex-cluded in 3/4% cases for CSfs, 5/6% for CSSPAIR, 4/5% for CSDixon and CSresoREF). CS strategies provided reliable aortic morphology measure-ments, as revealed by high correlation coefficients with reference (CSfs: r≥0.93; CSSPAIR: r≥0.93, CSDixon: r≥0.88, CSresoREF: r≥0.92) and low Bland-Altman biases, which further ranged within inter-operator variability (di-ameter≤0.55[-3.2;4.3]mm, length≤0.62[-14;15]mm, volume≤0.51[-9.1;10]ml). Overall, the best performance was obtained using CSfs strate-gy, with absolute differences below 15±12% compared to the reference. Conclusion: A non-contrast 3D aortic morphology MRI sequence combining CS, fat saturation and standard resolution achieved high agreement with conventional reference, while dividing average scan time by 3.