From 1974 through 1977 when our hospital mortality for aortic valve replacement and myocardial revascularization was 3.5% and 1.1%, respectively, hospital mortality for mitral valve replacement (MVR) was 8.3% (13/156)--as high as 14.9% in 1976. Transverse midventricular disruption (TMD) was present in 7 of 10 patients on whom an autopsy was done and was clinically diagnosed in 3 others without postmortem examination. Transverse midventricular disruption presented as refractory myocardial failure immediately on termination of bypass or later (1 to 5 days) after an initial period of good hemodynamics. It appeared to result when volume loading or afterload pressure was returned to the untethered ventricle after MVR performed with potassium-induced, cold cardioplegia and ischemic arrest. Operative techniques were modified to preserve a portion of the mitral suspensory mechanism, to extend the reperfusion interval following cardioplegia and ischemic arrest, and to control strictly ventricular volume and pressure loading following bypass. By utilizing these methods, TMD was avoided from 1978 through 1982, and hospital mortality for MVR was 3.7% (9/241). The improved hospital mortality and avoidance of TMD did not result from patient selection. Allowing adequate time for recovery of the myocardium after cardioplegia plus ischemic arrest prior to ventricular loading, preservation of mitral suspensory function, and strict control of preload and afterload pressures have been effective in lowering hospital mortality for MVR and have eliminated TMD in a 5-year period.