Left ventricular volumes and the ejection fraction can be determined by two-dimensional echocardiography. Because of methodological problems, cineventriculographic normal values can not be transferred to two-dimensional echocardiography. Therefore, normal values for two-dimensional echocardiographically determined left ventricular volumes and the ejection fraction were established in 55 normal controls. Basing on these volumes, the sensitivity and specificity of two-dimensional echocardiography for detection of impaired left ventricular function was analysed and compared with results from cineventriculography. Sensitivity for detection of a reduced ejection fraction was 84%, specificity 89% and predictive accuracy 98%. Thus, two-dimensional echocardiography can be used as a screening method for analysing left ventricular function. Analysis of beat-to-beat variation revealed that for two-dimensional echocardiography 3 consecutive beats should be analysed because evaluation of 5 instead of 3 beats increased reliability only by less than 1%. It was necessary to analyse the day-to-day variations. To separate a random error from a non-random error for an individual change, an absolute change of volumes of more than 10-12% and of ejection fraction of more than 10% had to be observed. Group changes can of course be smaller and yet significant. Intraobserver variation was not significant and below 5%. However, interobserver variability must be considered. This depends on observer experience. For a 4-chamber view no significant variability was observed, but for the RAO-equivalent view the observer variation was significant for end-diastolic and end-systolic volume. For the ejection fraction no significant difference was found. Two-dimensional echocardiography can be used for pharmacodynamic studies if percentage changes instead of absolute changes are listed.