The intraobserver, interobserver and beat to beat interpretative reproducibility of two-dimensional echocardiographic images of the mitral valve area has been studied retrospectively in a group of 37 patients affected by rheumatic mitral valve disease. Reproducibility has been expressed either in terms of mean absolute or percent error of duplicate measurements. A group of 11 normal subjects was used for comparison. In our normal group the intraobserver, interobserver and beat to beat reproducibility averaged 1.8 +/- 2.1%, 3.1 +/- 1.4%, 2.7 +/- 2.0% or 0.12 +/- 0.14 cm2, 0.21 +/- 0.10 cm2, 0.17 +/- 0.13 cm2, respectively. In our patient population the intraobserver, interobserver and beat to beat reproducibility averaged 2.7 +/- 2.7% 4.1 +/- 4.9%, 4.6 +/- 3.6% or 0.05 +/- 0.05 cm2, 0.08 +/- 0.10 cm2, 0.08 +/- 0.06 cm2. In both groups there was no statistical difference among intraobserver, interobserver and beat to beat reproducibility either in terms of percent or absolute value. The mean percent error did not significantly differ between normal or stenotic valves or, in this latter group, among valves of different sizes (less than or equal to 1.4 cm2; 1.5-2.4 cm2; greater than or equal to 2.5 cm2). The mean absolute error, on the contrary, statistically differed between the two groups or among valves of different sizes, being larger in normals or in valves greater than or equal to 1.5 cm2. Thus, the interpretative reproducibility for two-dimensional echocardiographic images of the mitral valve area is small and acceptable for most clinical purposes.(ABSTRACT TRUNCATED AT 250 WORDS)