Background/Objectives: Most data on radiofrequency (RF) effects come from ex vivo or in vitro studies that quantify lesions using width and/or depth, while electrophysiologists use manufacturers’ indirect indices. The objective of this study was to evaluate RF lesion volume by high-resolution MRI of excised lesions in an in vivo porcine model, comparing a low-energy long-duration (LE) (20 W, 50 s) RF application strategy with a high-energy short-duration (HE) (50 W, 20 s) one. Methods: Eighteen piglets were divided into LE (n = 9) and HE groups (n = 9). RF applications were performed at four locations in both atria. Animals were sacrificed after 5–7 days, and RF lesion specimens were excised, fixed, and analyzed by 9.4 Tesla MRI. RF lesion volume, variability (variance), depth, and any extracardiac lesions were compared between the groups. Results: Seventy RF applications were performed (36 LE, 34 HE). MRI analyzed 26 LE and 28 HE samples. The HE group showed 35% higher volume than the LE group (100.2 mm³ (±81.2) in LE vs. 178.3 mm³ (±163.7) in HE, p = 0.033). RF volume variance was 6.6 mm³ in LE and 40.3 mm³ in HE. The HE group had more complications (seven vs. zero, p = 0.02) and extracardiac lesions (18 vs. 14, p = 0.613). Conclusions: There was large and unpredictable variability in RF injury on the atrial wall, even under controlled conditions, which could explain arrhythmia recurrences. The greatest lesion variability was found during HE applications. The dose/effect relationship of RF needs careful study for treating cardiac arrhythmias.