Résumé : Objective: To prospectively compare diagnostic accuracy of fetal post-mortem whole-body MRI at 3-T vs. 1.5-T. Methods: Between 2012 and 2015, post-mortem MRI at 1.5-T and 3-T was performed in fetuses after miscarriage/stillbirth or termination. Clinical MRI diagnoses were assessed using a confidence diagnostic score and compared with classical autopsy to derive a diagnostic error score. The relation of diagnostic error for each organ group with gestational age was calculated and 1.5-T with 3-T was compared with accuracy analysis. Results: 135 fetuses at 12–41 weeks underwent post-mortem MRI (followed by conventional autopsy in 92 fetuses). For all organ groups except the brain, and for both modalities, the diagnostic error decreased with gestation (P < 0.0001). 3-T MRI diagnostic error was significantly lower than that of 1.5-T for all anatomic structures and organ groups, except the orbits and brain. This difference was maintained for fetuses <20 weeks gestation. Moreover, 3-T was associated with fewer non-diagnostic scans and greater concordance with classical autopsy than 1.5-T MRI, especially for the thorax, heart and abdomen in fetuses <20 weeks. Conclusion: Post-mortem fetal 3-T MRI improves confidence scores and overall accuracy compared with 1.5-T, mainly for the thorax, heart and abdomen of fetuses <20 weeks of gestation. Key Points: • In PM-MRI, diagnostic error using 3-T is lower than that with 1.5-T. • In PM-MRI, diagnostic scan rate is higher using 3-T than 1.5-T. • In PM-MRI, concordance with classical autopsy increases with 3-T. • PM-MRI using 3-T is particularly interesting for thoracic and abdominal organs. • PM-MRI using 3-T is particularly interesting for fetuses < 20 weeks’ gestation.