Introduction Bariatric surgery is effective in reversing adverse cardiac remodelling in obesity. However, it is unclear whether the three commonly performed operations; Roux-en-Y Gastric Bypass (RYGB), Laparoscopic Sleeve Gastrectomy (LSG) and Laparoscopic Adjustable Gastric Band (LAGB) are equal in their ability to reverse remodelling. Methods Fifty-eight patients underwent CMR to assess left ventricular mass (LVM), LV mass:volume ratio (LVMVR) and LV eccentricity index (LVei) before and after bariatric surgery (26 RYGB, 22 LSG and 10 LAGB), including 46 with short-term (median 251–273 days) and 43 with longer-term (median 983–1027 days) follow-up. Abdominal visceral adipose tissue (VAT) and epicardial adipose tissue (EAT) were also assessed. Results All three procedures resulted in significant decreases in excess body weight (48–70%). Percentage change in VAT and EAT was significantly greater following RYGB and LSG compared to LAGB at both timepoints (VAT:RYGB −47% and −57%, LSG −47% and −54%, LAGB −31% and −25%; EAT:RYGB −13% and −14%, LSG –16% and −19%, LAGB −5% and −5%). Patients undergoing LAGB, whilst having reduced LVM (−1% and −4%), had a smaller decrease at both short (RYGB: −8%, p < 0.005; LSG: −11%, p < 0.0001) and long (RYGB: −12%, p = 0.009; LSG: −13%, p < 0.0001) term timepoints. There was a significant decrease in LVMVR at the long-term timepoint following both RYGB (−7%, p = 0.006) and LSG (−7%, p = 0.021), but not LAGB (−2%, p = 0.912). LVei appeared to decrease at the long-term timepoint in those undergoing RYGB (−3%, p = 0.063) and LSG (−4%, p = 0.015), but not in those undergoing LAGB (1%, p = 0.857). In all patients, the change in LVM correlated with change in VAT ( r = 0.338, p = 0.0134), while the change in LVei correlated with change in EAT ( r = 0.437, p = 0.001). Conclusions RYGB and LSG appear to result in greater decreases in visceral adiposity, and greater reverse LV remodelling with larger reductions in LVM, concentric remodelling and pericardial restraint than LAGB.