Quantitative detection of left ventricular segmental wall motion abnormalities by any modality depends on the reference system used because of the dynamic geometry of contraction and cardiac motion within the thorax. To assess the feasibility and accuracy of quantitative analysis of left ventricular wall motion by two-dimensional echocardiography, we studied 61 subjects with the use of 44 different reference methods in each of three echocardiographic views: the parasternal short-axis view at the levels of the mitral valve and of the papillary muscles and an apical four-chamber view. The three major groups of reference systems used were those with a fixed external reference, a floating reference correcting for translation, and systems correcting for both translation and rotation. In the first part of this study the end-diastolic and end-systolic outlines of 20 normal subjects were stored in a computer and composite data of these 20 subjects were plotted to obtain a 95% confidence interval for measured normal fractional change for each reference method. In the second part of the project an additional prospective group of 10 normal subjects and a group of 31 "abnormal" patients had their left ventricular wall motion analyzed by similar methods and the results were compared with all the confidence intervals. One reference method was selected for each two-dimensional echocardiographic view based on the highest sensitivity and specificity found by statistical analysis; a floating-reference system including translation was found to be optimal for the apical four-chamber and parasternal short-axis views at the level of the mitral valve and a fixed external reference system was optimal for the short-axis view at the papillary level. The percent fractional shortening of radial dimensions (radial methods) and the percent fractional change in area measurements (area methods) during the cardiac cycle were also calculated at 5, 10, 20, 30 and 45 degree intervals around the outline perimeter for each subject according to each of the 44 different methods. Area methods yielded the same specificity and sensitivity as radial dimension analysis methods at 5 to 45 degree intervals. Ten normal subjects underwent repeat echocardiography within 2 days of their first study to examine day-to-day variation. Average change in mean contraction from day to day was 7% to 9% for radial methods and 9% to 13% for area methods. In conclusion, we present a computerized system for unbiased selection of optimal methods of analysis of left ventricular wall motion by two-dimensional echocardiography.(ABSTRACT TRUNCATED AT 400 WORDS)