The inter- and intraobserver variability in applying current diagnostic criteria for mitral valve prolapse as seen on two-dimensional echocardiograms is appreciable. It is therefore proposed that new methods employing computer-based pattern recognition algorithms be developed. The criteria will focus on the shape of the valve throughout systole. Shape analysis is first used to detect prolapsing sections of the valve. These prolapsing segments are fit using spline regression with a minimal number of knots. This mathematical description provides a basis for new shape descriptors.