Background-Ventricular restraint is a nontransplantation surgical treatment for heart failure. The effect of varying restraint level on left ventricular (LV) mechanics and remodeling is not known. We hypothesized that restraint level may affect therapy efficacy. Methods and Results-We studied the immediate effect of varying restraint levels in an ovine heart failure model. We then studied the long-term effect of restraint applied over a 2-month period. Restraint level was quantified by use of fluid-filled epicardial balloons placed around the ventricles and measurement of balloon luminal pressure at end diastole. At 4 different restraint levels (0, 3, 5, and 8 mm Hg), transmural myocardial pressure (P tm ) and indices of myocardial oxygen consumption (MV O 2 ) were determined in control (nϭ5) and ovine heart failure (nϭ5). Ventricular restraint therapy decreased P tm and MV O 2 , and improved mechanical efficiency. An optimal physiological restraint level of 3 mm Hg was identified to maximize improvement without an adverse affect on systemic hemodynamics. At this optimal level, end-diastolic P tm and MV O 2 indices decreased by 27% and 20%, respectively. The serial longitudinal effects of optimized ventricular restraint were then evaluated in ovine heart failure with (nϭ3) and without (nϭ3) restraint over 2 months. Optimized ventricular restraint prevented and reversed pathological LV dilatation (130Ϯ22 mL to 91Ϯ18 mL) and improved LV ejection fraction (27Ϯ3% to 43Ϯ5%). Measured restraint level decreased over time as the LV became smaller, and reverse remodeling slowed. Conclusions-Ventricular restraint level affects the degree of decrease in P tm , the degree of decrease in MV O 2 , and the rate of LV reverse remodeling. Periodic physiological adjustments of restraint level may be required for optimal restraint therapy efficacy.