In order to assess reproducibility of quantitative planimetry, three physicians trained in two-dimensional echocardiography performed five successive studies on one another over 2 weeks (30 total studies). Then each physician traced each study (90 total tracings) for left ventricular and atrial volumes and ejection fraction by means of a modification of Simpson's rule, and left ventricular mass and average wall thickness by means of a truncated ellipsoid formula. Calculation of intertechnician variability, intertracer variability, and 95% confidence limits showed that measurements of volumes were less reproducible than measurements of ejection fraction, average wall thickness, and mass. Mean intertracer variability of 15% exceeded mean intertechnician variability of 11%; this disparity was magnified in the subject who was technically difficult to image. Ninety-five percent confidence limits were: ejection fraction +/- 7%, average wall thickness +/- 9%, left ventricular mass +/- 12%, left ventricular end-diastolic volume +/- 11%, stroke volume +/- 14%, left ventricular end-systolic volume +/- 15%, and left atrial volume +/- 19%. Reproducible planimetry data can be obtained in normal hearts with the use of a protocol for quantitative imaging and planimetry.