The accuracy of measurements of mitral valve orifice area (MVA) from three-dimensional echocardiographic (3DE) image data sets obtained by a transthoracic or transesophageal rotational imaging probe was studied in 15 patients with native mitral stenosis. The smallest MVA was identified from a set of eight parallel short-axis cut planes of the mitral valve between the anulus and the tips of leaflets (paraplane echocardiography) and measured by planimetry. In addition, MVA was measured from the two-dimensional short-axis view (2DE). Values of MVA measured by 3DE and 2DE were compared with those calculated from Doppler pressure half-time (PHT) as a gold standard. Observer variabilities were studied for 3DE. MVA measured from PHT ranged between 0.55 and 3.19 cm2 (mean +/- SD 1.57 +/- 0.73 cm2), from 3DE between 0.83 and 3.23 cm2 (mean +/- SD 1.55 +/- 0.67 cm2), and from 2DE between 1.27 and 4.08 cm2 (mean +/- SD 1.9 +/- 0.7 cm2). The variability of intraobserver and interobserver measurements for 3DE measurements was not significantly different (p = 0.79 and p = 0.68, respectively); for interobserver variability, standard error of the estimate = 0.25. There was excellent correlation, close limits of agreement (mean difference +/- 2 SD), and nonsignificant differences between 3DE and PHT for MVA measurements (r = 0.98 [0.02 +/- 0.3] and p = 0.6), respectively. There was moderate correlation, wider limits of agreement, and significant difference between 2DE and PHT for MVA measurements (r = 0.89 [0.32 +/- 0.66] and p = 0.002), respectively. This may be related to the difficulties in visualization of the smallest orifice in precordial short-axis views. This study suggests that three-dimensional image data sets, by providing the possibility of "computer slicing" to generate equidistant parallel cross sections of the mitral valve independently from physically dictated ultrasonic windows, allow accurate and reproducible measurement of the MVA.