Four hundred M-mode echocardiographic surveys were distributed to determine interobserver variability in M-mode echocardiographic measurements. This was done with a view toward examining the need and determining the criteria for standardization of measurement. Each survey consisted of five M-mode echocardiograms with a calibration marker, measured by the survey participants anonymously. The echoes were judged of adequate quality for measurement of structures. Seventy-six of the 400 (19%) were returned, allowing comparison of interobserver variability as well as examination of the measurement criteria which were used. Mean measurements and percent uncertainty were derived for each structure for each criterion of measurement. For example, for the aorta, 33% of examiners measured the aorta as an outer/inner or leading edge dimension, and 20% measured it as an outer/outer dimension. The percent uncertainty for the measurement (1.97 SD divided by the mean) showed a mean of 13.8% for the 25 packets of five echoes measured using the former criteria and 24.2% using the latter criteria. For ventricular chamber and cavity measurements, almost one-half of the examiners used the peak of the QRS and one-half of the examiners used the onset of the QRS for determining end-diastole. Estimates of the percent of measurement uncertainty for the septum, posterior wall and left ventricular cavity dimension in this study were 10--25%. They were much higher (40--70%) for the right ventricular cavity and right ventricular anterior wall. The survey shows significant interobserver and interlaboratory variation in measurement when examining the same echoes and indicates a need for ongoing education, quality control and standardization of measurement criteria. Recommendations for new criteria for measurement of M-mode echocardiograms are offered.