A randomized prospective trial was undertaken to compare the hemodynamic performance and left ventricular regression after aortic valve replacement with the Toronto SPV stentless bioprosthesis and the Carpentier Edwards SAV (CE) bioprosthesis. Forty patients were randomized after the annular and sinotubular diameters had been measured. Early hemodynamic measurements were made with a thermodilution cardiac output catheter, and echocardiography was used thereafter. Left ventricular mass was assessed using magnetic resonance imaging (MRI) at 1 week, 6 months, and 32 months. The mean annular size was 25.3+/-2.2 mm (CE) and 25.5+/-1.5 mm (Toronto), although it was possible to implant valves with a mean diameter 3 mm larger in the stentless group (26.0+/-1.7 mm cf. 23.0+/-1.7 mm). Hemodynamic performance in the first 24 hours showed no significant difference between the groups, but there was a trend for shorter ventilation time and shorter stays in the intensive therapy unit in the stentless group. Echocardiography showed superior transvalvular gradients in the stentless group at 1 week (mean 5.5+/-3.1 mm Hg cf. 8.9+/-2.5 mm Hg), and this difference was maintained at a mean follow-up time of 32 months (3.5+/-0.6 mm Hg cf. 6.3+/-0.6 mm Hg). Similar regression of left ventricular mass was seen in both groups at 6 months, but at 32 months, measurement in diastole showed a reduction of 38% (P<.01) in the stentless group compared with 20% (P = ns) in the stented group, and measurements in systole showed a 23% (P<.01) and 13% (P = ns) reduction, respectively. This study confirms that a larger stentless valve can be implanted into a given size of aortic annulus with superior residual aortic valve gradients. These lower gradients seem to result in improved long-term regression of left ventricular mass as measured by MRI.