Transverse aortic constriction (TAC) is used as a model of left ventricular hypertrophy and failure; however, there is extensive variability in the hypertrophic response. In 43 mice that underwent TAC with a 7-0 polypropylene suture, 13 were identified by echocardiography with initial LV hypertrophy that halted or regressed over time. Post-mortem examination on 7 of these mice found the constricting band to be intact, but partially internalized into the aortic lumen, allowing blood flow around the stenosis. To confirm this prospectively in vivo we then followed 12 mice after TAC for 6 weeks, using a new high resolution 3D-MRI method to measure minimal aortic arch cross-sectional area (CSA). Three of the 12 mice developed a significantly increased aortic CSA (0.31 +/- 0.15 on day 2 vs. 1.11 +/- 0.29 mm2 on day 42; P < 0.05), which was independently confirmed by dissection. These mice had internalized part of the band within the aortic lumen and, by week 6, showed significantly less LV hypertrophy and better systolic function. Nine of the 12 mice showed no change in aortic CSA. Band internalization could be prevented when two banding sutures were placed side-by-side (n = 10). This is the first observation that a significant subset of animals following TAC bypass the stenosis resulting in partial regression of hypertrophy.