We studied the predictive accuracy and disparities among cineventriculographic ejection fraction, pre-ejection period over left ventricular ejection time (PEP/LVET) obtained from the systolic time intervals and the percent shortening of the internal echocardiographic diameter (% delta D) in assessing left ventricular performance in 453 consecutive patients without valvular heart disease. In 308 patients all three tests were normal, and in 78 patients all three tests were abnormal. Overall agreement (predictive accuracy) among ejection fraction (normal greater than or equal to 57), % delta D (normal greater than or equal to 28%) and PEP/LVET (normal less than or equal to 0.42) was 85%. In 67 patients disparities among the tests as measures of global left ventricular performance were found. The major mechanisms accounting for such disparities were: (a) large segmental contraction abnormalities which selectively distort the % delta D and ejection fraction and (b) diminished isovolumic pressure (less than 45 mmHg) which distorts PEP/LVET. When patients with segmental contraction abnormalities and low isovolumic pressure were excluded the agreement between PEP/LVET and ejection fraction was 97%, ejection fraction and % delta D 98% and PEP/LVET and % delta D 97%. The combined uses of systolic time intervals and echocardiogram minimizes error due to segmental contraction abnormalities and isovolumic pressure. If both PEP/LVET and % delta D are concordant the agreement with ejection fraction is 94% for normal and 99% for abnormal left ventricular function.